Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Transfer Of The Partial Reinforcement Extinction Effect Across Tasks In Normal And Retarded Boys
(USC Thesis Other)
Transfer Of The Partial Reinforcement Extinction Effect Across Tasks In Normal And Retarded Boys
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
ThU dUsertation haa been
microfilmed exactly aa received ® 7-6488
a
BARRON, Robert Francis, 1929-
TRANSFER OF THE PARTIAL REINFORCEMENT
EXTINCTION EFFECT ACROSS TASKS IN NORMAL
AND RETARDED BOYS.
University of Southern California, Ph.D., 1967
Psychology, clinical
University Microfilms. Inc.. Ann Arbor, Michigan
TRANSFER OF THE PARTIAL REINFORCEMENT
EXTINCTION EFFECT ACROSS TASKS
IN NORMAL AND RETARDED BOYS
by
Robert Francis Barron
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Psychology)
February 1967
UNIVERSITY OF SOU TH ERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
L 0 8 ANGELES, CALIFORNIA 9 0 0 0 7
This dissertation, written by
under the direction of his Dissertation C om
mittee, and approved by all its members, has
been presented to and accepted by the Graduate
School, in partial fulfillment of requirements
for the degree of
Date Ee.bx.uajr.y.,..!9.6.7........................
DISSERTATION COMMITTEE
D O C T O R O F P H I L O S O P H Y
Dean
Chairman
ACKNOWLEDGMENT
I wish to express my appreciation to the members of
my committee: Dr. Alfred Jacobs, Dr. William W. Grings,
Dr. L. Douglas DeNike, and Dr. Robert B. McIntyre for their
interest and assistance in this research. I am particularly
indebted to my chairman, Dr. Jacobs, for his guidance, sug
gestion and encouragement and to Dr. Norman Cliff for his
help with some difficult statistical problems. I wish also
to thank the administrators and staff members of the Man
hattan Beach City Schools, the Inglewood Unified School
District, and the Redondo Beach City Schools for their
cooperation and Dr. Arthur Silverstein of Pomona State Hos
pital for his help in the pilot study which preceded the
final study.
TABLE OF CONTENTS
Page
ACKNOWLEDGMENT.................................. ii
LIST OF TABLES.................................. v
LIST OF ILLUSTRATIONS........................... vii
Chapter
I. INTRODUCTION ............................. 1
II. REVIEW OF THE LITERATURE................. 10
Two-Factor Theories of Frustration
Research Findings in Amsel's Theory
Partial Reinforcement
Cognitive Dissonance Explanation of PRE
Persistence as the PRE in Human Studies
Amsel's Theory at the Human Level
III. DEVELOPMENT OF THE PROBLEM............... 36
Specific Predictions of Test Performance
IV. METHODOLOGY............................. 45
Subjects
Apparatus
Procedure
Measurement of Response
Testing Conditions
V. RESULTS................................. 61
Adjustment in Criterion Scores
Hypothesis 1
Hypothesis 2
Hypothesis 3
Hypothesis 4
Chronological Age and Performance
Interview Analysis
iii
Chapter
Page
VI. DISCUSSION 84
Frustration Effect
Anticipatory Frustration Responses
Transfer of the PRE
Persistence or Perseveration
Frustration Tolerance
Motor Performance
The Interview and Behavioral Observations
VII. SUMMARY
REFERENCES 111
APPENDIXES 120
APPENDIX A: PARENTAL CONSENT LETTER
APPENDIX B: STATISTICAL TABLES . .
121
123
LIST OF TABLES
Table Page
1. Experimental Design ....................... 52
2. Mean Number of Trials to Extinction on
Seguin Form Board...................... 70
A. Mean Time per Trial in Seconds for Performance
of Section Two of Task I— Adjusted Means for
Retardates in Parentheses ............... 124
B. Intelligence Groups X Treatments Grouping
Analysis of Variance of Trial 3 of Section
Two on Task I .......................... 125
C. Trend Analysis of Trials 4 Through 12 of
Section Two on Task I ................... 126
D. Trend Analysis of Trials 4 Through 12 of
Section Two on Task I for Normal Boys . . . 127
E. Trend Analysis of Trials 4 Through 12 of
Section Two on Task I for Retarded Boys . . 127
F. Mean Time per Trial in Seconds for Performance
of Section One of Task I— Adjusted Means for
Retardates in Parentheses ............... 128
G. Intelligence Groups X Treatments Grouping
Analysis of Variance of Trial 3 of Section
One on Task I
H. Trend Analysis of Trials 4 Through 12 of
Section One on Task I ................... 130
I. Trend Analysis of Trials 4 Through 12 of
Section One on Task I for Normal Boys . . . 131
J. Trend Analysis of Trials 4 Through 12 of
Section One on Task I for Retarded Boys . . 131
Page
K. Analysis of Variance in A 2 X 2 X 2 Factorial
Design of Trials to Extinction on Seguin
Form Board after Log Transformation . . . .
L. Mean Time per Trial in Seconds for First
Eight Trials on Seguin Form Board— Adjusted
Means for Retardates in Parentheses . . . .
M. Correlation Coefficients for the Relationship
of Chronological Age with Task Performance .
132
133
134
vi
LIST OF ILLUSTRATIONS
Figure Page
1. Task One Apparatus................... 48
2. Task Two: Seguin Form Board Adaptation ... 50
3. Performance Speed in Section Two of
Turning Test...................... 64
4. Performance Speed in Section One of
Turning Test...................... 67
5. Performance Speed on Seguin Form Board .... 75
CHAPTER I
INTRODUCTION
This study investigated the role of frustrative non-
reward on boys1 performance of a psychomotor task and the
effect such nonreward had on the persistence on a second
psychomotor task. Amsel1s frustration theory (1962) pre
dicts that the occurrence of nonreward when reward is
anticipated leads to frustration which increases drive level
and intensifies ongoing responses. This theory is an exten
sion of Hull's learning theory (1943, 1952) which emphasizes
the active role of nonreward in both the intensification and
inhibition of a response. The development of Amsel's theory
and his construct of the anticipatory frustration response
(r^) has led to an explanation of persistence of nonrewarded
behavior as attributable to the partial reinforcement ex
tinction effect (PRE). The present study investigated both
the immediate effect of nonreward upon psychomotor perform
ance and the transfer effect of nonreward experience from a
first to a second psychomotor task. The study was concerned
with these effects on the performance of boys of both normal
and retarded intellectual status.
2
! In a review of contemporary development in learning
theory, Berlyne (1964) refers to Amsel's contribution when
he states, "This line of thought has been radically trans
forming the study of such basic phenomena as extinction,
partial reinforcement and discrimination" (p. 122).
Amsel's recent investigations have been concerned
with the role of frustrative nonreward in discrimination
I
learning (Amsel and Ward, 1965). However, the first studies
that appeared in the literature were rather simply concerned
with the motivational properties of nonreward on instru
mental behavior which had previously been continuously
rewarded (Amsel and Roussel, 1952). Amsel classified such
an operation as "frustrating" and defined frustration as "a
conceptualization of a hypothetical, implicit reaction
elicited by nonreward after a number of prior rewards"
(Amsel, 1958, p. 103).
Under such conditions of prior reward, the absence
of reward results in a frustration response (Rp) which is
*'a primary, aversive, motivational condition" (p. 103) and
has three major consequences: (1) an increase in drive
level which is demonstrated in the energizing of ongoing
behavior and is called the frustration effect (FE), (2) a
stimulus state of frustration (Sp) and, in later trials
(3) the development of anticipatory frustration responses
and their consequent stimuli (rf”sf mechanism).
j Virtually all of Amsel's studies have been conducted
i
at the animal level and the bulk of research pertaining to
the theory has also been at the infrahuman level. Amsel
remarked at an early point in the theory's development that
his was a very restricted use of the concept of frustration
and that his formulation was less involved with the more
emotional or clinical connotations (1958, p. 103). In a
later and more complete presentation of the theory he sug
gested that his concepts might serve as a "learning theory
counterpart of frustration tolerance" (1962, p. 315) and
added it might be possible to recast the Rosenzweig theory
or the Rotter theory of frustration into a Hullian frame
work. Certainly, his use of the term frustration for
responses following nonreward and his use of the word per
sistence for resistance to extinction have implications for
i
behavior at the human level.
There have been a number of applications of Amsel's
theory to the human level. Longstreth (1960) and Penny
(I960) were the first investigators to study Amsel's theory
Using young children as subjects. More recently there have
been studies of Amsel's formulations with mentally retarded
children (Longstreth, 1965; Semler, 1965). However, when
retarded children have been used as subjects, the purpose
has not been to contrast frustration behavior of normal with
retarded children since no single study has included both
groups. After a brief review of the literature on frustra-
[tion and mental retardation, Heber (1964) indicated that
j
despite the common assumption that retardates are more
susceptible to frustration there are remarkably few studies
in this area and little evidence to support the assumption
that retardates respond any differently to frustration than
normals. He concludes that the generalization that retar
dates have less frustration tolerance than normals has sel
dom been put to the test.
Amsel has frequently used the two-stage straight
runway to investigate FE. This apparatus consists of a
straight runway leading to a mid-goal box followed by a
second straight runway leading to a final goal box. After
number of trials in which the animal receives a food reward
in both goal boxes, nonreward occurs occasionally in the
toid-goal box while reward always occurs in the final goal
i
box. This results in increased speed of running in the
second runway on those trials in which nonreward occurs in
the mid-goal box. Amsel and others have found this effect
in a number of studies and, as an empirical finding, it is
po longer seriously challenged. Amsel interprets the in
creased speed of running following nonreward as FE due to
increased drive resulting from a R„.
r
Hull (1952) theorized that when an animal runs in a
single runway for a number of continuously rewarded trials
fractional anticipatory goal responses and their consequent
stimuli (r^-Sg) develop, becoming more frequent the closer
(the animal is to the goal response (R_) . Amsel has stated
that a similar process occurs for frustration responses when
nonreward occurs after rg-sg ^as developed. When nonreward
trials are interspersed with rewarded trials then not only
will Rj, occur on the early nonrewarded trials but on subse
quent nonrewarded trials there will be the development of
anticipatory frustration responses and their consequent
stimuli (rf“sf)•
Running the alley will then elicit both r -s and
g g
rf-sf, confronting the animal with both approach and avoid
ance tendencies. As a result of such conflict stemming from
partial reinforcement, acquisition speeds are slower for the
partially reinforced (PR) animal in the early stages of
i
acquisition than the continuously reinforced (CR) animal.
i
The greater resistance to extinction of the running response
in the PR animal which is called the partial reinforcement
Extinction effect (PRE) can be explained by this theory.
The PR animal has learned to run the alley during acquisi
tion in the presence of anticipatory frustration while the
CR animal has not. In extinction the PR animal continues to
run longer than the CR animal because r^-s^ has been con
ditioned to r -s for the PR animal but not the CR animal.
g g
This, Amsel's explanation, is only one of many
explanations for the PRE that has been offered in the past
30 years (Jenkins and Stanley, 1950; Lewis, 1960), but it is
one that is currently receiving much attention. Recently
[investigators (Jenkins, 1962; Theios, 1962; Ross, 1964) have
shown that the PRE can be sustained through a block of rein
forced trials, a finding which is compatible with Amsel's
theory but contrary to some other explanations. Working
within the Amsel framework and at the animal level Ross
(1964) was also able to show that the PRE produced by one
task transferred to a second task.
The present study investigated the PE with normal
and retarded boys on a task designed to be analogous to
Amsel's two-stage runway and also investigated the transfer
of the PRE from a first to a second task. Both tasks were
adaptations of relatively simple psychomotor tests. Part
One of the study was concerned with the introduction of non
reward and the observation of both FE and the effect of
r^-s^. Task I involved an adaptation of the Minnesota Rate
of Manipulation Test. Part Two was concerned with the
influence of the preceding experience of reinforcement on
persistence in a new task. Task II involved the use of the
iSeguin Form Board.
In Part One the Minnesota test was so modified that
it consisted of only two straight segments each containing
eight discs and the form board into which they fit. Comple
tion of Segment One was indicated by pressing a button
switch and completion of Segment Two by pulling a lever
switch. Marble rewards could be received at the midpoint
and the end of each trial. The performance required was the
Turning Test of the Minnesota test. Task I is thus analo
gous to the alley-goalbox-alley-goalbox apparatus used in
animal studies. Half of the retarded boys and half of the
normal boys received 50 per cent reinforcement at the mid
goal box (button response) while the remaining subjects
received 100 per cent reinforcement. All subjects received
100 per cent reinforcement at the final goal box (switch
response).
In Part Two, the Seguin Form Board was connected to
a large box upon which was mounted both a switch and a but
ton in such a way that the experimenter could make either
available to the subject. At the base of this box was an
opening through which marbles could be dispensed. Half of
I
each of the four groups pressed the button at the completion
of the Seguin, half pulled the switch. All subjects were
instructed to play the game until they wished to quit but
only the first four trials were rewarded.
Within the context of Amsel's theory, the following
hypotheses were made:
Part One
1. Performance on the second half of a task will be
more rapid after intermittent frustrative nonreward at the
midpoint than after continuous reward at the midpoint.
2. Performance on the first half of a task will be
slower following intermittent frustrative nonreward at the
Imidpoint of the task than after continuous reward at the
I
imidpoint.
Part Two
3. Intermittent frustrative nonreward on a first
motor task will lead to greater resistance to extinction
(persistence in the face of continuous nonreward) on a
second motor task than will continuous reward on the first
motor task.
4. Frustrative nonreward on a second motor task
will lead to greater increase in performance speed of imme
diately subsequent performance on this task if the experi
ence on the first motor task has been of continuous reward
rather than intermittent frustrative nonreward.
No specific hypothesis was made concerning the dif
ference between normals and retarded subjects and in this
!
respect the study might be considered an approach to the
i
problem of individual differences which Cantor (1963) has
called the "gross comparison" method. The study did offer
l
some opportunity to investigate the general assumption that
i
retardates are less tolerant of frustration and less per
sistent in the face of failure. However, because of many
alternate predictions that might be made due to possible
differences between retarded and normal subjects such as
expectancy of reward, emotionality, and speed of learning,
it seemed that a first study of Amsel's theory with both
9
I
(normals and retardates might be more prudent to observe than
i
predict.
CHAPTER II
REVIEW OF THE LITERATURE
One of the earliest theoretical formulations of
frustration defined it as a state resulting from goal inter
ference (Dollard, Doob, Miller, Mowrer, and Sears, 1939).
Yet frustration has been conceived so broadly as to be con
sidered a basic element of all learning. Spence (1960)
reformulated Hull's concept of inhibition and referred to
the new formulation as "frustration phenomena" which occur
because of a failure of reward and which contribute to the
extinction of a response in instrumental reward learning.
Some authors have referred to the period of time between the
j
Conditioned stimulus and the unconditioned stimulus in
j
classical aversive conditioning as "frustrating" to the sub
ject. One of the most inclusive definitions has been pro
posed by Sheffield and Campbell (1954) in which frustration
is related to any consummatory response arousal that is not
I
immediately completed.
From time to time major reviews of the concept of
frustration appear in the literature and attempts are made
to gain organization and perspective in the field. Twenty-
five years ago Britt and Janus (1940) indicated that frus-
10
! 11
i
{tration can be viewed in three ways: (1) as a conflictual
situation, (2) as a state of the organism, and (3) as a reac
tion to this state. These authors suggested that the frus
tration process includes all three conceptualizations. In a
tnore recent and extensive review of the literature Lawson
and Marx (1958) stated that no one had listed clearly the
antecedent and consequent conditions that are pertinent to a
study of frustration. In his book, Frustration and Conflict
(1962), the English author Yates has provided a similar
review of frustration theories devoting a considerable
amount of space to Maier's fixation theory which has re
ceived much criticism in this country. Neither the Lawson
and Marx article nor the Yates book give an extensive treat
ment of Amsel's frustration theory although both classify it
under the category of the "two-factor frustration theories"
i
i
jalong with the Brown and Farber (1951) conceptualization of
frustration. Actually, Amsel's major theoretical papers
have only appeared within the past eight years (Amsel, 1958,
1L962) . The present review will be primarily concerned with
these two-factor theories and the contemporary literature.
i
Two-Factor Theories of Frustration
Marx (1956) identified three different formulations
of frustration within the basic framework of S-R learning
theory. All three conceptualized frustration as an internal
emotional state that has at least two consequences, (1) it
increases drive and thus energizes ongoing or immediately
'subsequent behavior, and (2) produces distinctive stimuli to
which responses might become attached.
One statement of this position was by Child and
Waterhouse (1952) who identified frustration as a state pro
duced by blocking a goal response. This state increased
drive as well as elicited interfering responses. The Child
|and Waterhouse position has not been developed to the point
that it would have the status of a well-defined theory.
However, the role of competing responses will again be per
tinent in the later discussion of partial reinforcement.
Brown and Farber (1951) published a theoretical
article concerned with the concept of emotion as an inter
vening variable in dynamogenic as contrasted to associative
behavior theories. Frustration is identified as an emo
tional state typically produced by blocking a goal response.
Irhis condition might occur by obstructing goal behavior, by
(delaying reward or denying reward. All of these operations
function "in some way to arouse an incompatible reaction
tendency" (p. 481). In Hullian terms, "Frustration is the
(
consequence of either (1) the simultaneous activation of two
t
competing excitatory tendencies, or (2) the presence of a
single excitatory tendency and an opposing inhibitory ten
dency" (p. 481). The authors then propose that such frus
tration may (1) increase the general level of motivation and
(2) produce unique internal stimuli (Sp) . It will be noted
13
Jthat the theory makes little distinction between conflict
and frustration.
Brown and Farber proposed that Sp might be the mech
anism to account for the appearance of similarity of "frus
tration" responses in the second of two different situations
and that such "stimulus elements could mediate the transfer
of these responses from one situation to a dissimilar one
. . (p. 479). Bernstein's animal study (1957) suggested
that Sp might account for some demonstrated commonality in
responses from a first to a second frustration condition.
Amsel's theory is quite similar to the Brown and
Farber formulation but with one important addition. Actu
ally, Amsel's theory contains three factors. Frustration
I
not only adds an increment to drive and produces its own
distinctive stimuli, it also leads to the development of
I
r^-s^ in later trials. It is this third factor of rf~Sf
which has received the greatest attention in the recent work
of Amsel and his colleagues.
Amsel's theory concerns itself only with nonreward
as the frustration producing operation. In a recent experi
ment on the effect of frustrative blocking on motivation,
Uyeno (1965) refers to Amsel's frustrative nonreward
hypothesis and Brown and Farber's frustrative interruption
hypothesis and concludes, "It is suggested that in both
studies the expectation of the animal seems to be an essen
tial psychological variable arousing frustration" (p. 208).
' * 14
I Research Findings in Amsel's Theory
Amsel's early research (Amsel and Roussel, 1952)
dealt with the first factor, the withdrawal of reward and
the demonstration of the consequent motivational effect in
terms of intensification of ongoing or subsequent behavior.
There are many studies at both the animal and human level
that offer evidence that interruption, blocking, delaying or
hot rewarding a goal seeking response will result in the
intensification of the immediately subsequent behavior.
Working with young children Screven (1954), Screven and
Cummings (1955), Haner and Brown (1955), Longstreth (1960),
Holton (1961), Ryan and Cantor (1962) and, Cantor and Ryan
(1964) have all found evidence of increased vigor after
interrupting or not rewarding a goal response. However,
there are many frustration theories which predict such a
finding.
The second factor of the two-factor theories, Sp,
received some attention in Amsel's earlier experimentation.
{Amsel and Ward (1954) did attempt to link responses to Sp
but the learning was highly unstable and a more parsimonious
interpretation of their findings was possible. Aside from
the Bernstein study mentioned earlier there are few studies
in this area.
The third factor of Amsel's theory, anticipatory
frustration, is its most distinguishing characteristic and
it is this construct that is used to explain the active role
I 15
jof nonreward in inhibition of performance, partial rein-
I
i
forcement effects in acquisition and extinction, and dis
crimination learning. These responses are a variant of
Hull's fractional anticipatory goal responses (r -s ).
9 9
Amsel feels that Hull's term should be the generic term for
three types of anticipatory responses— rewarding (rrs),
punishing (r s), and frustrating (r-s) .
F x
While Brown and Farber suggested that Sp, the frus
tration drive stimulus, might mediate similar frustration
responses across a variety of frustration situations, the
stimulus evoked by the anticipatory frustration response
(s^) suggests a more precise mechanism for such transfer.
A particularly crucial evaluation of the role of s^ in the
i
PRE can be seen in a recent study by Ross (1964). Working
at the animal level, Ross trained rats to one of three dif
ferent responses, half of each of the three groups received
partial reinforcement, half received 100 per cent reinforce
ment. According to Amsel's theory only the partially rein
forced groups would experience r^-s^. Thus, for only these
groups would the sf be associated with the goal response.
Following this training, motivation was changed (from hunger
to thirst) and a new task was presented with 100 per cent
reinforcement for all groups. In the extinction trials of
this new task, those groups of rats with the history of par
tial reinforcement showed greater resistance to extinction
than those rats with a history of continuous reinforcement,
16
but only when the goal response of the second task was com
patible with the original goal response.
Ross's explanation is that only the partially rein
forced groups has sf attached to a goal response by virtue
of the first training task and that it is only for these
groups that a learned response will be elicited by the sf
which occurs for all groups in the extinction trials of the
second task. Aside from demonstrating the power of s^ to
transfer responses across tasks, Ross also showed that this
power was not lost by a change in motivation, by an inter
vening period of continuous reinforcement, nor by a change
in the goal response.
Many of the contemporary experimental studies that
i
have investigated Amsel's formulations have concerned them-
i
Iselves with factors effecting the magnitude of FE or the
i
PRE. For example, Greenbaum and Levy (1965) found Amsel's
FE relatively uneffected by the presence of multiple stress
or the administration of thorazine. McHose and Ludvigson
(1964) and Badia (1965) investigated the role of drive and
!FE. In a situation analogous to the double runway, Daven
port and Thompson (1965) found evidence for the Amsel FE in
monkeys. Peckham and Amsel (1964) found the FE effected by
the magnitude of the reward in a within subjects design.
Studies concerned with the PRE have studied the role of goal
box cues (Elmes, 1964; Kirkpatrick, Pavlik and Reynolds,
1964), the number of reinforced trials (Madison, 1964),
jlatent extinction (Dyal, 1965), interpolated continuous
reinforcement (Perry and Moore, 1965; Ross, 1964) and goal
box behavior (Longstreth, 1964). All of these studies have
offered support for Amsel's formulations and despite alter
nate theories, the frustration hypothesis must certainly be
considered the major contemporary theoretical explanation
for partial reinforcement effects.
Partial Reinforcement
There have been two major reviews of the partial
reinforcement phenomena. In 1950 Jenkins and Stanley pre
sented a review and critique of the literature from 1939 to
1949. Ten years later Lewis (1960) reviewed the literature
for the decade beginning in 1950. By this time the litera
ture was so voluminous that Lewis reports only the partial
^reinforcement phenomena in the experimental designs using
discrete trials rather than the "free responding" designs
used in Skinnerian type studies.
Jenkins and Stanley presented three empirical gen
eralizations after reviewing over fifty studies, (1) learn
ing during acquisition proceeds somewhat more rapidly under
continuous rather than partial reinforcement, (2) perform
ance tends to be lower under partial rather than continuous
reinforcement, and (3) resistance to extinction is greater
after partial rather than continuous reinforcement "when
behavior strength is measured in terms of single responses"
(p. 222). Lewis remarks that after nine more years of
18
experimental research the empirical generalization of
igreater resistance to extinction following partial reinforce-
i
ment still stands. However, he also states there is empiri
cal evidence to show an interaction between percentage of
reinforcement and number of trials during acquisition. On
early trials continuously reinforced groups perform better
but during later trials the partially reinforced groups sur
pass them. This is a prediction that Amsel's theory makes
but there is no mention of this in the Lewis article.
A simple S-R learning theory does not predict the
PRE since it states only that a reward strengthens a re
sponse and absence of a reward weakens the response.
Humphreys' studies (1939) with intermittent reinforcement in
i
eyelid conditioning questioned such a straightforward role
jof reinforcement.
Jenkins and Stanley classified the theoretical
Explanations of partial reinforcement phenomena into two
major categories corresponding to the traditional classifi-
i
pation of learning theories, S-R theory and cognitive
l
jtheory. Within the framework of S-R theory are four types
of explanation. Skinner (1938) offered the "response unit"
explanation in which a single reward after a series of unre
inforced responses gives the whole group of responses the
status of a single response. Mowrer and Jones (1945) also
proposed the "response unit" explanation as well as a second
possibility, the discrimination hypothesis. These authors
19
admit the discrimination hypothesis might be considered a
more precise form of Humphreys' expectation theory. In
essence, the discrimination hypothesis states that acquisi
tion trials and extinction trials are more similar in their
stimuli for the partially reinforced and therefore more dif
ficult to discriminate.
The Hull-Sheffield explanation was in terms of
aftereffects (Sheffield, 1949). The partially reinforced
subject is conditioned to the aftereffects of nonreinforce
ment while the continuously reinforced subject is not. The
explanation predicts that the PRE should only occur if
acquisition trials are massed.
Denny (1946) proposed that secondary reinforcement
was the responsible agent for differences between partial
jand continuously reinforced animals' behavior. However,
I
Denny was primarily concerned with differences in rate of
learning between these two groups and did not even find dif
ferences in resistance to extinction. Hulse and Stanley
(1956) were not able to replicate Denny's findings and sug
gested that secondary reinforcement could not explain all
i
the effects of partial reinforcement.
Humphreys' expectancy theory was the only explana
tion which Jenkins and Stanley classified as a cognitive
approach. In essence, the explanation states that subjects
who receive partial reward in training take longer to real
ize that reward will not occur in extinction. Partial reward
I 20
ihas trained such subjects to expect a certain amount of non-
j
reward and hence they fail to discriminate acquisition from
extinction trials as rapidly as do continuously reinforced
subjects.
Two more recent explanations could be classified
under S-R theory. The competing response theory has been
proposed by Weinstock (1954) and in a somewhat different
form by Hulse and Stanley (1956). In Weinstock's approach,
the partially reinforced make competing responses on nonre
warded trials during acquisition but these responses "habit
uate" and therefore leave the partially reinforced animal
more free of such responses during extinction. Hulse and
Stanley found that secondary reinforcement could explain
some but not all of the partial reinforcement phenomena and
offered a theory based on competing responses. According to
this explanation the PR animal learns to "do something else"
(p. 226) on nonrewarded trials and does this something else
during extinction and thus preserves the "conditioned eating
response" from quick extinction.
Lewis classifies both Amsel's and Logan's theory
(1960) as explanations based on mediating responses but both
theories are treated very briefly. Presumably the publica
tion date of Lewis' article was responsible for the limited
attention to both Amsel and Logan. Logan's formulation
seems to have received little attention or research aside
from his own studies. Like Amsel's position, Logan's
21
incentive theory uses the r -s mechanism but without any
g g
construct of frustration. The partially rewarded animal and
the continuously rewarded animal both develop these antici
patory goal responses which in later trials act as incentive
to the goal response. However, during acquisition the PR
i
animal may learn to inhibit goal responses in the presence
jof nonreward without incentive being greatly diminished. In
the extinction trials the CR animal inhibits goal responses
for the first time and the incentive motivation of r -s
g g
diminishes more quickly than for the PR animal. The PR ani
mal has already learned to inhibit goal responses in the
absence of reward with incentive motivation of r -s remain-
g g
fing relatively strong. Like Amsel, Logan has also applied
i
his theory to discrimination learning (Brown and Logan,
1965).
Before discussing a new cognitive explanation some
bomments should be made about the durability of the older
theories. The Humphreys' expectancy theory is challenged by
i
the studies that show the PRE can be sustained through a
tlock of reinforced trials (Jenkins, 1962; Theios, 1962;
Ross, 1964). The Hull-Sheffield aftereffects theory has
been seriously challenged a number of times and specifically
by a study which could not replicate Sheffield's original
findings (Wilson, Weiss and Amsel, 1955). In fact, any
study such as Weinstock' s (1954) that obtains the PRE on
trials spaced as far apart as 24 hours is counterevidence
for the aftereffects theory. Both Longstreth (1964) and
|McCoy and Marx (1965) have found no support for Weinstock's
interference-habituation theory. The response unit explana
tion does not seem applicable to situations involving dis
crete, spaced trials.
Since 1960 a new cognitive theory of the PRE has
been proposed, Festinger's cognitive dissonance explanation
of partial reinforcement (Festinger, 1961; Lawrence and
Festinger, 1962).
Cognitive Dissonance Explanation of PRE
Festinger first spoke of the psychological effects
of insufficient rewards in 1961 and a year later (Lawrence
i
^nd Festinger, 1962) published a more complete explanation
of the adaptation of cognitive dissonance theory to the par-
I
jtial reinforcement phenomena at the animal level. Cognitive
idissonance is a motivational state which arises when "an
I
individual's actions are not entirely consistent with what
he knows about the environment" (p. 33). The individual is
jnotivated to reduce such dissonance. At the human level
jthis happens when an individual expends much effort for a
reward he recognizes as meager. Left in such a state of
cognitive dissonance he must reduce this dissonance by
changing his behavior (no longer working for the reward) or
change his cognitive information about the reward. In the
latter case his recourse may be to increase the value of the
23
reward, or the activity, in his own cognitions. The activ
ity or reward may thus develop "added attractions."
Mindful of the reduced cognitive capacities of the
rat these authors hold that such an animal may have "items
bf information" which can be related to each other in con
sonant and dissonant ways. The total motivational picture
is the weighted sum of task relevant consonant and dissonant
Relationships. Concerning the partial reinforcement situa
tion when nonreward occurs in acquisition trials dissonance
is created and the animal cannot reduce this by not running
since this would create greater dissonance because of his
information that rewards do occasionally occur. Thus disso
nance is reduced by "extra attractions" development.
t
In extinction, the control group will have their
t
first experience with dissonance and to some extent extra
I
I
attractions will develop on the nonrewarded trials. "Such a
!
process is however, relatively slow and gradual" (p. 56).
Dissonance is now more effectively reduced by not running.
For the partially reinforced animal, the extra attractions
|are already present at the time of extinction trials. Both
groups experience dissonance in extinction but the magnitude
is greater for the control group.
There follows a report of 16 animal studies which
explore some of the deductions from such a theory. In
experiment 7 Lawrence and Festinger challenge Amsel's
theory. Using an apparatus similar to Amsel's two-stage
24
J
jrunway but with mid and end goal boxes being distinct, one
t
1
group of rats never received reward in the mid-goal box
although like the PR and CR groups they were always rewarded
in the final goal box. This group was called the "Zero
reward group." Wagner (1959) ran a similar group in a two-
stage runway and found no FE in the second runway which was
support for Amsel's position that FE only occurs to nonre
ward after some anticipation has been developed for reward.
However, Festinger and Lawrence were only concerned with
resistance to extinction and the extinction trials were only
in the first runway. They found the zero reward group
showed greater resistance to extinction than the PR or CR
groups. Since there was no measure of running speed in the
second runway, nothing could be said about FE. The authors
conclude that Amsel cannot explain both FE and PRE by the
i
same mechanism of frustration since according to Wagner no
FE occurs for the zero reward mid-goal box group yet accord
ing to Lawrence and Festinger this group is the most resist
ant to extinction.
i
While this admittedly does raise some problems for
the frustration hypothesis it seems also to raise equally
difficult problems for cognitive theory. It would seem that
Lawrence and Festinger would now have to explain why cogni
tive dissonance ever arose at the midpoint of a continuously
rewarded task and why there ever was a need for this mid
goal box to develop any extra attractions at all.
25
I
i
j One final remark should be made about cognitive dis-
i
sonance theory. In testing for the "added attractions" that
developed in the goal box of PR animals (experiment 1) ,
Lawrence and Festinger attempted to see if the PR animals
would learn a new task to a now empty goal box faster than
the CR animals. Half of each of these groups also learned
this new task to a new but empty goal box as a control mea
sure. The results indicated that the PR group also learned
faster than the CR group to the new goal box which the
authors could only attribute to "some generalization of the
partial reward effect" (p. 66) . Experiment 4 had a similar
finding. However, no further explanation is offered and it
yould seem to demand some explanation of how "extra attrac
tions" developed in one goal box as a result of dissonance
reduction could transfer to a new and different goal box.
Persistence as the PRE in Human Studies
In a recent review Feather (1962) classified per
sistence studies as being of three major classes. In the
I
i
first class persistence is treated as a general personality
trait which transcends momentary situational influences.
Early characterlogical studies and later factor analytical
studies have looked for a common trait across a variety of
i
physical and mental tasks which were concerned with endur
ance and persistence in these tasks. The studies do suggest
the existence of a persistence trait with a low order
26
[positive correlation with intelligence and introversive ten-
j
'dencies.
In the second classification are the studies which
identify persistence as resistance to extinction. Such an
approach has the major emphasis upon situational variables
and tends to exclude personality variables. In this class
iof studies persistence is typically related to the partial
reinforcement effect and the repeated performance on an
activity in the face of consistent nonreward.
In the third class of studies persistence is treated
as a motivational phenomenon and investigates such personal
ity factors as need achievement (n ach) interacting with
such situational factors as task difficulty. Older children
j
and adults are usually the subjects in such studies because
i n ach is usually measured by verbal projection techniques,
and the trait itself is not present very early in life.
It is the second class of studies, those which
identify persistence with resistance to extinction, which
jare pertinent to the present study. Within this framework
jalmost all studies which investigate the PRE at the human
level can be classified as persistence studies. In fact,
Feather begins his review of this class of studies with
reference to Humphreys' eyelid conditioning studies of the
late 1930's. It will be remembered that Humphreys' work was
mentioned by Jenkins and Stanley as one of the first studies
of partial reinforcement.
27
A study by Grosslight and Child (1947) titled,
|
"Persistence as a function of previous experience of failure
followed by success," is not mentioned by Feather but is
referred to by Lawson and Marx as a frustration study. The
study has some significance to the present one since re
tarded subjects were used. Grosslight and Child had re
tarded subjects pull a series of nine handles to obtain a
piece of candy. One group received a piece of candy on each
of the ten trials, a second group had one nonrewarded and
nine rewarded trials and a third group had two nonrewarded
and eight rewarded trials. Because of the human capacity
for verbal mediating responses, the authors reasoned that a
single failure followed by success would cause the greatest
increase in persistence during extinction and a greater pro
portion of failures during training would produce relatively
small increment to persistence. The authors found some sup
port for their position although the group with two nonre
warded trials tended to persist less than the group with
Only one nonreward experience. Unfortunately there was no
I
formal control group and the rationale for using institu
tional retardates was not given.
While the Screven and Cummings study (1955) is not
technically a study of persistence, the authors were con
cerned with the effect of continuous nonreward on the per
formance of children turning a crank for a marble reward.
These authors suggest that eventually nonreward increases
28
1
the range or variation of response strength within single
trials and the means of these trials suggesting over-all
faster or slower crank turning than rewarded groups is some
thing of an artifact.
Lewis and Duncan (1958) studied the function of per
centage of reinforcement and number of acquisition trials on
jthe persistence of college students in playing an electronic
slot machine and found that the smaller the percentage of
reinforced trials the greater the persistence and that the
larger the number of acquisition trials the lower the per
sistence. An earlier study (Lewis and Duncan, 1957) had
indicated that magnitude of reward was also positively
related to persistence.
James and Rotter (1958) studied the role of a rather
Uniquely human factor as it might effect the partial rein-
I
forcement phenomena. Working within the framework of a
cognitive learning theory, the authors suggested that a
factor influencing persistence during extinction trials will
be the speed and manner in which the subject "recategorizes
the situation" once extinction trials begin. The subject
who believes that rewards are contingent upon his perform
ance skill (internal control) will behave differently than
the subject who believes that rewards are contingent upon
luck (external control).
It is reasoned that when subjects are so instructed
as to believe rewards are contingent upon skill they will
29
j
[take longer to re-evaluate the situation after the change
from training to extinction trials than subjects who have
been instructed to believe rewards are contingent upon sheer
chance. It is only among those subjects who believe that
rewards are contingent upon chance, that the PRE will be
i
obtained. With "chance" instructions continuously rein
forced subjects would quickly recategorize the situation
[when extinction trials begin and would soon stop responding.
However, such subjects would continue to respond longer dur
ing extinction following partial reinforcement due to what
the authors call the "gambler's fallacy." These subjects
have learned during training that bad luck was followed by
good luck and respond during extinction under the same
hypothesis. Under "skill" instructions continuously rein
forced subjects might even persist longer than partially
reinforced subjects. Giving college students a task under
"skill" or "chance" instructions resulted in support for the
authors' hypotheses. Actually, this explanation of the PRE
jis quite similar to Humphreys' explanation based on "expect
ancy" with the reservation that it only applies when sub
jects believe reward is contingent upon external factors.
Working with young children and within the framework
pf Amsel's theory, Longstreth (1960) predicted that "the
stronger the symbolic expectative response of reward, the
greater the amount of frustration when that reward is
removed" (p. 667). This implies that the stronger the
I 30
expectation of reward, the greater the frustration and the
!
less persistence will be shown at the introduction of nonre
ward. Children who had learned to associate reward with a
light cue did persist less in a lever holding response dur
ing extinction than children who had not developed this
association.
Amsel's Theory at the Human Level
This final section will be concerned with adapta
tions of Amsel's theory to human performance. Certainly the
major support for the frustration hypothesis comes from ani
mal studies. The Longstreth study (1960) was one of the
first studies of Amsel's theory with children.
A distinction is sometimes made in the literature
between a "within" and a "between" effect of frustrative
nonreward. A within effect concerns differences in perform
ance of the partially rewarded subjects on performances
immediately following nonreward as compared to performances
immediately following reward. A between effect concerns
i
differences in performance between the partially rewarded
subjects and a control group of subjects who are continu
ously rewarded.
Many studies have used a sequence of two lever push
ing or pulling tasks which might be considered the counter
part of mid-goal box and final goal box in an attempt to
produce a situation at the human level which is analogous
to Amsel's two-stage runway.
31
! Penny (1960) used such an apparatus with kinder-
i
garten children and studied the effect of nonreward at two
levels of habit strength. For the experimental groups non
reward occurred only on the first of the two levers and
measurements were taken of the start and movement times on
the second lever. The control groups received continuous
reward on both levers. The FE was only obtained with the
high habit group.
Penny reported a within FE in which speed of move
ment on the second lever was faster following nonreward on
the first lever than on trials following reward on the first
lever. A between FE was found with the high habit groups
but only on four of the twelve nonrewarded trials. Penny
concluded, "Under the experimental conditions of the present
study the effect of nonreinforcement is not stable but rela
tively transient" (p. 213).
In a very recent study Ryan (1965) also used the two
successive levers design with kindergarten children in a
ktudy of incentive value and Amsel's FE. As in the Penny
Study both starting speeds and movement speeds were recorded
for each lever. The experimental groups received 50 per
cent reward on the first lever, the control groups received
100 per cent reward. All groups received 100 per cent on
the second lever. A between groups FE was found for both
measures but evidence for a within group FE was weak.
Incentive value was not a significant factor in performance.
Ryan suggests that the within group frustration effect
|
Which has been so consistently found in animal studies is
difficult to obtain with children. He refers to an early
unpublished study in which he failed to replicate Penny's
findings in obtaining a within group FE and suggests that
Penny's findings may have been an artifact of the training
techniques used for the high habit groups in that study.
An unusual finding in Ryan's study was the increased
starting and movement speeds for the 50 per cent groups on
the first lever. In a study with kindergarten children
using only a single lever Bruning (1964) had also found a
50 per cent reinforcement group performing with greater
speeds during acquisition than a 100 per cent reinforcement
group. However, a general finding in animal studies has
been poorer acquisition performance for the PR group, at
least initially. In an animal study with an Amsel-like two-
stage runway Holder, Marx, Holder and Collier (1957) had
found frustration in the mid-goal box produced the FE in
Jrunway two but slower speeds in runway one. This is a find
ing that is predictable from Amsel's theory. Other chil
dren's studies using a single lever response (Ryan and
Cantor, 1962; Cantor and Ryan, 1964) have also suggested
faster acquisition performance for partially reinforced
groups.
However, the single lever response does not separate
FE from anticipatory frustration especially if the inter
33
trial interval is short. Even the successive lever task may
not be too comparable to the two-stage runway.
Semler (1965) used Amsel's theory to study persist
ence in severely retarded institutionalized children. It
was hypothesized that such individuals would have reliable
individual differences in persistence across similar tasks
jwhen all subjects received the same number of rewarded
trials in acquisition series. A significant correlation was
found between resistance to extinction on a crank turning
task and a lever depressing task both of which gave M&M
candies on a partial reinforcement basis during acquisition.
Semler accepts this as an indication there are relatively
stable differences in persistence among subjects and accord
ing to Amsel's theory interprets this to mean that high per-
sistors have a capacity to resolve the approach-avoidance
conflict of r and r^ more rapidly than low persistors.
After obtaining a reliable persistence index on all subjects
he predicts that on the subsequent learning of a position
response high persistors will choose a 50 per cent rein-
jforced stimulus over a 0 per cent reinforced stimulus more
frequently in early trials than low persistors. His study
showed some support for this prediction.
Semler's study is, of course, correlational and, as
a persistence study, would be in the class of general trait
studies that Feather mentions even though the index is
resistance to extinction. It might be disputed that Semler
34
i
!
!has demonstrated persistence across tasks because the two
I
tasks used to establish the persistence index are almost
identical. Even though there was no experimental manipula
tion of nonreward this does not rule out the possibility
that learning occurred and that situational factors did
influence performance across tasks.
A final study to be mentioned is one in which there
jwas experimental manipulation of reward on earlier tasks.
Steigman and Stevenson (1960) investigated the influence of
reinforcement schedule on three early tasks to learning on a
fourth task. One group of nursery school children experi
enced considerable nonreward on the trials of three differ
ent tasks while a second group received considerable reward.
When presented with a difficult discrimination problem, sig
nificantly more of the "reward" group solved the problem
while the "nonreward" group made significantly more per-
iseverative errors. The authors interpret the study in terms
of increased drive created by the earlier frustration for
the "nonreward" group as being responsible for disorganiza
tion in the later complex learning task. The authors also
refer to some unpublished work which suggested that if the
criterion task was simple the performance of the "nonreward"
group was not inferior.
This last study is not actually a persistence study
nor do the authors make any reference to Amsel's constructs.
However, it does have a pertinence to the present study
jsince it investigates the role of nonreward on early tasks
j
to performance on a later task.
CHAPTER III
DEVELOPMENT OF THE PROBLEM
Amsel's extention of Hull's theory has resulted in
very explicit explanations of the phenomena of frustration,
persistence, and discrimination learning. Yet, as the last
chapter revealed, the support for this theory as well as the
support for most of S-R learning theory comes from research
at the infrahuman level. The generalization of such find
ings to the human level is always a highly tentative one and
not until such research is actually performed at the human
level can the power of such a theory be established. Yet,
if research at the human level is done with artificial or
animal appropriate tasks, support for the theoretical
hypotheses may contribute very little to the prediction of
human behavior.
In the research with children reviewed at the end of
the last chapter, the pulling of a lever was the typical
task demand. But apart from such research situations, rela
tively little of human conduct has to do with lever pulling.
In the studies cited in which mentally retarded persons were
used as subjects, none of these studies were concerned with
differences between retardates and normals. Retardates were
36
37
used either because they were a more available population
or because as a group they do present a certain combination
of attributes. Mentally retarded children are more limited
in their symbolic processes than normal children of the same
age. Some investigators may wish to work with human sub
jects who have very limited symbolic processes yet are
physically mature enough to perform certain motor responses.
In some ways such research tends to minimize the most sig
nificant differences between the animal and the human sub
ject. The sum and substance of such research may be to
establish that under certain carefully restricted conditions
the immature human organism behaves similarly as does the
rat under parallel conditions.
However, the application of theories supported by
observations at one level of the phylogenetic scale fre
quently need translation if they are to be applied at a dif
ferent level. In the last chapter the adaptation of cogni
tive dissonance theory from the human level to the animal
ilevel was reviewed. When Hullian type theories are studied
at the human level some translations may also be needed and
such may be the case with Amsel's central construct of
anticipatory frustration responses.
As mentioned earlier this construct of Amsel's is a
variant of Hull's fractional anticipatory goal responses.
Hull postulated that as an animal such as the rat has re
peated experiences running an alley for a food reward the
38
I
lanimal may begin to anticipate this reward while running in
the alley and make fractional or incipient parts (r^s) of
the goal response. In fact in his last publication Hull
(1952) had a short section on the r -s mechanism in which
9 9
he talks of the "realization of an anticipation and its
frustration" (p. 133). Spence (1956) has suggested that the
stimuli (SgS) resulting from such responses may act as in
centive and increase motivation. It has also been suggested
that at the human level, such anticipatory responses might
be treated as "expectancies" (Brown, 1961).
Working with normal kindergarten children, Long-
streth studied (1960) a similar relationship between sym
bolic (verbal) expectative responses and frustration as the
relationship between fractional anticipatory responses and
frustration. Thus with human subjects, regular reinforce
ment of an instrumental response might lead to anticipations
Of reward and the production of both peripheral anticipatory
responses and verbal expectative responses. According to
jAmsel, nonreward would only be frustrating after some expec
tation of reward had been established and with repeated
experiences of nonreward would lead to a frustration expec
tancy. In fact, Amsel (1962) refers to the S-R explanations
of the partially reinforcement effect as "conditioning
expectancy" as contrasted to the "cognitive expectancy"
approach of Festinger.
39
If Amsel’s model is investigated at the human level,
i
it is clear that the theory concerns itself only with rela
tively immediate situational factors affecting frustration
and persistence and there is little reference to the influ
ence of general traits. The Semler study (1965) was an
exception in this respect in that it did use Amsel's con
struct to investigate stable traits of persistence. Semler
correlated scores representing individual differences in
persistence on two similar tasks during extinction trials
with performance on a later learning task involving nonre
ward. The significant correlation between persistence on
the first two tasks was taken as an indication of the
presence of a relatively stable trait. Semler used mentally
retarded subjects with an IQ range of 24 to 58 and reported
a nonsignificant correlation of .20 between intelligence and
persistence. It is possible that the truncated range of the
retarded population may have limited the magnitude of the
correlation.
However, to use an experimental approach to the
study of persistence in which the focus of attention in upon
situational factors does not rule out the possibility that
certain general traits might be present. Amsel (1958) had
hinted at the possible role of relatively stable reactions
to frustration when he mentioned the possibility of his
formulation being the S-R learning theory counterpart of
Rosenzweig's frustration tolerance. At another point, Amsel
40
also suggested that in the adult organism some response
|
bight already be connected to s^ "which determines that
organism's characteristic reaction to anticipated frustra
tion" (1958, p. 111).
Yet it still seems possible that the degree of
reward or success in one task can effect performance in an
immediately following task and Amsel's theory is appropriate
for the study of the factors that might operate in such
transfer. There certainly are real life situations for the
child in which one task follows another, each offering
opportunities for reward and nonreward. The classroom is
one of these situations, the psychological testing situation
is another. Even in the administration of a psychometric
instrument such as the Wechsler Intelligence Scale for
Children, the child is given a series of tasks each starting
I
with success and being terminated after a series of fail
ures. It seems very likely that persistence in final tasks
given might be influenced by expectancies established on
1
early tasks.
j The present study is concerned with the role of
frustrative nonreward on performance on a psychomotor test
and the influence such nonreward experiences might have on a
second psychomotor test. The design of such a study would
call for the adaptation and modification of a standard test
in such a way that it would be analogous to the double run
ways used in studying FE in rats. In a pilot study involving
41
{approximately 20 moderately mentally retarded boys, various
i
!
adaptations of the Minnesota Rate of Manipulation Test were
used. The most successful adaptation appeared to be the use
of two straight line segments of the Turning Test. The seg
ments were arranged spatially in such a way as to be similar
Ito the consecutive alleys of the animal studies and a marble
dispensing machine was placed at the end of each segment.
The subject's task was to turn over each of eight
discs in each segment starting at the left and working in
the direction of the marble dispensing machine. At the end
of Segment One he pushed a button to obtain a possible
marble reward and at the end of Segment Two he pulled a
switch to obtain a possible marble reward. The apparatus
and task suggest an even closer parallel to Amsel's double
ialley than do the consecutive levers used in the children's
studies reviewed earlier. The major experimental manipula
tion was partial versus continuous reinforcement at the end
of Segment One. This allowed for the observation of the FE
Jin Segment Two and the effects of anticipatory frustration
in Segment One as a consequent of nonreward.
The second task chosen was the Seguin Form Board.
(In the pilot study a reaction time task to a light stimulus
had been attempted as the second task but the retardates had
difficulty in attending to the light and this task was dis
carded.) The Seguin task is attractive to most boys and the
performance requirements are so self-apparent that little
! 42
i
jinstruction is needed. It consists of ten geometric forms
I
and a formboard in which they are to be placed. The form-
board was attached to a marble dispensing machine and upon
completion of the task the subject either pushed a button or
pulled a switch for a possible marble reward.
With all subjects receiving the same amount of con
tinuous reward followed by continuous nonreward, the task on
the Seguin allowed for a measure of the transfer of the
effects of partial reinforcement from the preceding task.
Pushing a button was an identical response to the partially
reinforced response in the first task and different from
pulling a switch which was the continuously reinforced re
sponse. It was predicted that the transfer of the partial
reinforcement effect would be greater for those subjects who
were directed to signal their completion of the Seguin task
by button than by switch.
The pilot study used only institutionalized male
retardates and its primary purpose was to establish appro
priate tasks and instructions for the proposed study. The
primary study was planned to use boys with normal intelli
gence attending regular classes in school and boys with mild
retardation who attend classes for the educable mentally
retarded (EMR classes) . In essence, the same study was done
twice, once with a normal group and once with a mildly re
tarded group.
43
j To the extent that conscious symbolic (verbal)
i
processes may play a significant part in the older child's
performance of even simple psychomotor tasks, a brief inter
view consisting of approximately twelve questions was de
signed to be administered at the end of the testing of each
subject. The interview was constructed in such a way as to
assess the subject's verbal report of the immediately pre
ceding experiences as well as to study the extent to which
these verbal reports corresponded to actual performance and
to theoretical predictions. The interview was discarded in
the pilot study after it was readily apparent that the
institutional retardates did not have the verbal facility
ifor such a task. However, the interview was used in the
major study.
Specific Predictions of
Test Performance
I
Part One
1. Task performance on Section Two of Task I will
I
be more rapid for subjects who receive frustrative nonreward
experience at the end of Section One than for subjects who
receive continuous reward at the end of Section One.
2. Task performance on Section One of Task I will
be slower for subjects who receive frustrative nonreward
experience at the end of Section One than for subjects who
receive continuous reward at the end of Section One.
44
I
I
I P art Two
I
1 3. Subjects who have had frustrative nonreward on
Task I will show greater persistence on Task II than sub
jects who have had continuous reward on Task I. The per
sistence will be even greater when the button is used in
Task II than when the switch is used.
4. Subjects who have been continuously reinforced
on Task I will perform faster on Task II following nonreward
experience on Task II than will subjects who have had inter
mittent frustrative nonreward on Task I.
CHAPTER IV
METHODOLOGY AND DESIGN
i
Subjects
The sample consisted of 56 Caucasian boys obtained
from the Manhattan Beach City Schools, the Inglewood Unified
School District and the Redondo Beach City Schools. The
three school systems are geographically adjacent to each
Other and are within the Los Angeles County area. Six indi
vidual schools within these systems supplied the total 56
subjects. All schools are in a middle or upper middle class
socio-economic area. Half of the subjects were enrolled in
regular classes in public schools and were between 10 years
1
2 months and 14 years 3 months of age. Half were enrolled
in EMR classes in public school and were between 10 years 3
months and 14 years and 2 months of age.
The author met with the school officials of the
jthree school systems and the nature of the study was pre
sented. However, individual teachers were not informed of
the purpose or nature of the study. While the school files
on individual students were not made available to the author,
the credentialed school psychologist in each of the three
districts cooperated in the selection of subjects. The
46
I
Jauthor, who served as the sole experimenter, assigned sub-
|
jects to the treatment conditions.
A request was made to the school psychologist to
select boys in EMR classes in the age range between 10 and
14 years and to obtain a boy of normal intelligence in regu
lar classes matched in age for each retarded boy. The aver
age age for each of the two groups obtained was identical,
12 years 6 months. The 28 boys in regular classes had
normal or above normal IQ scores on group tests periodically
given by the school systems. Twenty-five of these boys had
IQ scores in the 90 to 110 range, two had scores in the very
superior range and one in the dull normal range.
Assignment to EMR classes is typically made on the
i
basis of academic achievement and scores on individually
administered intelligence tests. Of the 28 boys selected
j
from EMR classes, 27 had IQ scores in the 60 to 75 range,
one had an IQ score of 55. Children who were identified in
the school records as severely emotionally disturbed or
physically handicapped were not considered as eligible for
jthe study. The study also excluded students who lived in
institutional settings even though they might attend public
schools in the community. Two subjects, both in the normal
group, were replaced, one because of apparatus failure dur
ing testing and one because of misassignment to treatment
conditions.
47
i
! The two school systems from which 53 of the 56 sub
jects were obtained insisted that parental consent be given
for a child to participate in the study. A copy of the
letter sent to parents is included in Appendix A. No par
ents contacted objected to their child's participation.
Apparatus
Task I
The first task consisted of a modification and adap
tation of the Minnesota Rate of Manipulation Test (Super,
1949). The present study used only one row of the forraboard
consisting of 16 holes and 16 discs (see Figure 1). The row
was divided into two equal, 18 1/2-inch segments of 8 holes
and discs each and the subject was required to perform only
the Turning Test. An apparatus was designed to house each
lof these segments as well as two marble dispensing machines
and timing devices. The apparatus for this first task was
4 feet long and hinged in the middle to be folded and port
able. At the end of Segment One of the formboard was a
6 1/2-inch high box-like structure that housed and hid a
marble dispensing machine. On top of this structure was an
ivory colored electric button switch protruding from a rec
tangular black metal plate. Directly below the button at
the base of the structure and on a level with the formboard
was an opening and a small tray to receive marbles. At the
end of Segment Two a similar box-like structure occurred
Ivory button on black plate
0 0 ft Q 0 Q 0 g { " j ' h » g | r Q 0 0 0 ^ Q
l -^9
§E°SBi??iS£St«
Marble openings and trays
Figure 1
TASK ONE APPARATUS
4*
00
(which also hid a marble dispensing machine. On the front
side of this structure was a dark brown colored electric
lever switch which protruded from a rectangular white metal
plate. Directly below this plate was an opening and tray to
receive marbles identical to the opening and tray at the
base of the button switch. Two toy gum ball machines were
modified in such a way to dispense marbles one at a time
when operated manually by the experimenter. They were at
all times hidden from the view of the subjects. The ivory
button (hereafter called the button) and the brown lever
(hereafter called the switch) were so connected to a
Standard Electric timer that the operation of either could
be set to start or stop the timer. Marble rewards were all
of the same size and color (red crystal) throughout the
study.
Task II
The second task was a minor adaptation of the Seguin
Form Board (Arthur, 1930). The Form Board used in the study
was a 13 1/2-inch by 18 1/2-inch board with 10 recessed
areas into which could be placed 10 blocks of varied geo
metrical form. There is only one correct recess for each of
the 10 forms. The adaptation of the test for this study
(see Figure 2) consisted of adding a box-like structure to
the right edge of the board. This structure housed and hid
a marble dispensing machine identical to those used in
Task I. At the base, facing the subject was an opening and
r
/ « £ > ^ ^ r - s \
■
/ O g " A y — V \
< ^ > E 5 \
Figure 2
TASK TWO: SEGUIN FORM BOARD ADAPTATION
Double hinged flap covering button
Rotating box with switch exposed
Marble opening and tray
51
!
ia small tray to receive marbles. On top of this structure
was a "lazy susan" unit to which was attached a second
square box. This box contained an ivory button (top) and a
brown lever (side) with cover plates identical to those used
in the first task. A double-hinged flap made it possible to
conceal the button when the switch was exposed or to conceal
the switch when the button was exposed. The lazy susan made
it possible for the experimenter to make either switch or
button available to the subject in the same orientation
which subject had had to the switch and button of the first
task. As in the first task, both switch and button were so
connected to the timer that either could stop or start the
instrument.
Procedure
Although the study had two subject groups, intellec
tually normal and mildly retarded boys, the major independ
ent variable was the manipulation of reward at the end of
the first segment of Task I. The design is therefore a
study of nonreward and persistence with normal boys and a
simultaneous replication of the study with mildly retarded
boys. The experimental conditions are presented in Table 1.
All subjects received 12 trials on Task I. A single
trial consisted of completion of both segments of the Turn
ing Test. Half of the normal group and half of the retarded
group were reinforced only 50 per cent of the time at the
completion of Segment One (button response) while the
52
TABLE 1
EXPERIMENTAL DESIGN
Group
Task I: Minnesota Rate of
Manipulation
Task II:
Seguin Form Board
Section One
(Button
Response)
Section Two
(Switch
Response)
Reward Nonreward
Trials Trials
1-4 5-54
Normal
Intermittent
Frustrative
Nonreward
Continuous
Reward
Switch Response
Button Response
Continuous
Reward
Continuous
Reward
Switch Response
Button Response
Retarded
Intermittent
Frustrative
Nonreward
Continuous
Reward
Switch Response
Button Response
Continuous
Reward
Continuous
Reward
Switch Response
Button Response
53
jremaining subjects received 100 per cent reinforcement after
j
completion of Segment One. All subjects received 100 per
cent reinforcement at the completion of Segment Two (switch
response). On Task II half of the subjects who had partial
reward on Task I and half of the subjects who had continuous
reward on Task I signaled their completion of the Seguin
Form Board by the button response, the remaining subjects
signaled their completion by the switch response. The con
ditions of reward were the same for all subjects on Task II,
the first four trials were rewarded, all subsequent trials
were nonrewarded.
Age was an important control in the assignment to
treatments because of the known correlation between age and
psychomotor performance. Prior to any contact of the
experimenter with the subjects, the schools provided the
I
experimenter with only the name, age, and classification as
to normal or retarded for each subject available. For each
group, normal and retarded, the subjects were then arranged
fay the experimenter in an order of increasing age and
assigned alternately to the conditions of partial or contin
uous reward for Section One of Task I. The partially rein
forced treatment groups and the continuously reinforced
treatment groups were then each assigned alternately to
switch or button response on Task II. This resulted in
eight groups equal in chronological age.
54
■ All testing and interviewing was done by the same
experimenter, the author of the study. Each boy was seen
individually, on the school premises during regular class
t
hours. Aside from the instructions for the tasks and the
interview questions, all conversations were kept at a minimum
without making the situation formidable. Order of testing
depended primarily on the school schedule and the availabil
ity of the subject; however, an attempt was made to test
partial and continuous subjects alternately and retarded and
normal subjects alternately. The latter task was difficult
since EMR classes frequently have different hours than regu-
lar classes.
Task I
When each boy came to the testing room, the experi
menter shook hands with him, introduced himself and said,
"(boy's first name), you are going to play some games to win
some marbles. If you get enough marbles you will win a
prize. The boy who wins the most marbles will get the big-
i
gest prize." The Task I apparatus was in full front view on
I
l a desk when the boy entered the room. He was led to a chair
in front of the apparatus and the experimenter continued
with,
"The marbles will come out here after you press the
button and here after you pull the switch" (experi
menter demonstrates task as he speaks). "When I say
'go' you turn over the circles in the first board one
at a time like this then push the button here. If
you work well, you may get a marble. When I say 'go'
again you turn over the circles in the second board
55
! the same way then pull the switch down. If you work
well you may get a marble."
The experimenter gave the subject a small glass dish
in which he could place the marbles he won and then seated
himself behind the apparatus, facing the subject, and said,
"Remember the boy who gets the most marbles wins the biggest
prize so you do well to get as many marbles as you can."
The experimenter prepared the subject with the
remark "Ready" about two seconds before saying "Go." When
the experimenter said "Go" he started the electric timer
with a hand switch. The subject's depression of the button
stopped the timer and the experimenter covertly released a
marble from the gum ball machine.
i
The time taken for the segment was recorded, the
timer reset and the button switch depressed by the experi-
j
jnenter to again open the circuit. The subject was then
j
given the "Ready, go" signal for Segment Two. The time
elapse between segments was approximately 10 seconds for all
subjects. As the subject pulled down the switch at the end
j
pf Segment Two, the experimenter released a marble from the
i
I
Second gum ball machine and recorded the time for that seg
ment. The entire sequence constituted one trial. The time
interval between trials was approximately 30 seconds. Dur
ing this time the experimenter pulled up the switch to open
the circuit, reset the timer and reset the loading lever on
each of the gum ball machines. Each subject received 12
trials. The partially rewarded group received no marbles at
56
i
I
jthe end of Segment One on trials 4, 6, 7, 9, 10, and 11, but
i
always received marbles at the end of Segment Two. Reset
ting the loading lever on the gum ball machines between
trials always gave rise to the sound of marbles being moved
in the machine. However, the same sounds were made even if
the machine were set not to deliver a marble.
Task II
After the completion of trial 12 on Task I, the
experimenter said, "Now we will play a different game to win
more marbles, it will only take me a few minutes to set it
up." The subject was given the glass cup containing the
marbles he had won to hold while the experimenter removed
the apparatus for Task I and placed it on the floor at the
right of the subject. The Seguin Form Board apparatus was
placed in front of the subject with the forms correctly in
place and either the button or switch exposed as was
appropriate for the specific subject.
The forms were removed from the board by the experi
menter and stacked in three piles at the rear of the board
!in the same arrangement as the test is used in the Arthur
Point Scale of Performance Tests, Form I (Arthur, 1930).
The experimenter then said:
"When I say 'Go' put the pieces back in the board one
at a time as well as you can and then push the button
(pull the switch) and you may get a marble. Which
hand do you write with?" (After subject indicates pre
ferred hand) "O.K. just use that hand. Remember when
I say 'Go' use only one hand and put the pieces back
then push the button (pull the switch) . The boy who
57
\ wins the most marbles will get the biggest prize."
j (The experimenter paused and thus increased the sali
ence of his last remark.) "You can quit playing this
game when you want to— you tell me when you want to
stop."
The task demands for the Seguin are so obvious that
no demonstration was necessary. The time interval between
Task I and Task II was approximately 3 minutes, while the
time interval between trials on the Seguin was the same as
the interval between trials on Task I, approximately 30 sec
onds. In this interval the experimenter would record the
time for each trial, reset the timer, open the circuit by
depressing the response button or pulling up the switch and
reset the loading lever on the gum ball machine.
All subjects received marbles on the first four
trials of the Seguin, no marbles were received from the
i
fifth trial onwards. After the tenth nonrewarded trial,
I
I
the experimenter again said, "you can quit playing this game
when you want to, you tell me when you want to stop."
Thereafter, the experimenter repeated the same statement
lafter every fifth nonrewarded trial. The experimenter indi
cated that the task was over after trial 54 if the subject
had not quit before this.
As soon as the subject indicated that he would quit,
the experimenter said "O.K. I'd just like to ask you a few
questions before you leave." The following questions were
asked and the answers recorded by the experimenter.
58
! 1. (Pointing to the Seguin) Do you think you maybe
could have won more marbles? (If the subject
answered no, he was asked if he thought it was
"possible" to win more marbles.)
2. (If the answer was yes to first question) Fine,
how could you have won more marbles?
3. Why do you think you didn't always get marbles?
4. Which game did you like most to do?
5. Which game did you like least to do?
6. If you came back tomorrow which one game would
you like to play?
7. If you wanted to win more marbles which game would
you play?
8. (Revealing unexposed button or switch on Seguin
apparatus) Do you think if you had used this but
ton (or switch) you would have
a. played the game more times. Why?
b. quit the game sooner. Why?
c. played the game the same number of times
you did.
I'll say that again (entire question with alter
natives repeated)
9. Which one, the switch or the button, work better
or are they both the same? (Experimenter points
to the button and switch on the Seguin.)
10. (The experimenter then moves to the apparatus of
Task I and directs the subject's attention to
this.) On this part, the first half, you did it
twelve times, just before we stopped doing this,
i did you think you were doing it
a. faster. Why?
b. slower. Why?
c. or the same as other times.
I'll say that again (entire question and alterna
tives repeated)
The following two questions were asked only of those
subjects who received partial reinforcement on Task I.
11. When you didn't get a marble here sometimes what
did it make you do?
12. On this part, the second half (pointing to Sec
tion Two), when you didn't get a marble here
(pointing to button response) do you think you
worked
a. faster. Why?
b. slower. Why?
c. the same as when you got a marble here.
I'll say that again (entire question and alterna
tives repeated)
No subject was given the opportunity to count the
i
total number of marbles he received. The experimenter took
the glass cup saying he would make a record of how many the
boy had received but that he would have to wait until others
did the games before it could be determined who got the most
marbles. However, each boy was told he did well and was
jgiven a token gift, a ten-cent compass, for "playing the
games."
Measurement of Response
All time measures were obtained on a Standard Elec-
i
trie timer sensitive to one-hundredth of a second. A small
six-volt light was attached to this circuit in such a manner
that it would indicate to the experimenter when the clutch
jwas engaged and the switch and button circuits were open.
The experimenter transcribed time measures after
i
each measured response and also transcribed answers to the
I
interview questions as well as anecdotal comments on the
subject's behavior, remarks and general demeanor during the
!
Session.
.
Testing Conditions
Each boy was tested individually during regular
classroom time. Testing was always in a room with only the
experimenter and the subject present. The room was usually
an office room in the school but in one instance it was an
empty classroom. Both the subject and the experimenter were
seated throughout the task performances. Testing time was
lapproximately 45 minutes for each subject.
CHAPTER V
RESULTS
The experimental predictions are restated below:
Part One
1. Task performance on Section Two of Task I will
be more rapid for subjects who receive frustrative nonreward
experience at the end of Section One than for subjects who
receive continuous reward at the end of Section One.
2. Task performance on Section One of Task I will
be slower for subjects who receive frustrative nonreward
experience at the end of Section One than for subjects who
i
receive continuous reward at the end of Section One.
j?art Two
3. Subjects who have had frustrative nonreward on
Task I will show greater persistence on Task II than sub
jects who have had continuous reward on Task I. The per
sistence will be even greater when the button is used in
Task II than when the switch is used.
4. Subjects who have been continuously reinforced
in Task I will perform faster on Task II following nonreward
62
{experience on Task II than will subjects who have had inter-
i
jmittent frustrative nonreward on Task I.
Adjustment in Criterion Scores
The major index for tests of Hypotheses 1 and 2 is
that of speed of performance measured in hundredths of a
second. It was apparent from first inspection of the data
i
that while the initial levels of performance on the first
i
three trials of the partial reward group and the continuous
reward group of normal boys were similar, the partially
rewarded retarded boys performed at slower speeds on both
the Turning Test and the Seguin Form Board than did the con
tinuously rewarded retarded boys prior to any experimental
(manipulation of reward. For Task I it was possible to
[derive from the first three trials (all rewarded) regression
I
{coefficients of speed of performance in these trials on Sec
tion One and speed of performance in the same trials on Sec
tion Two. The correlation coefficients between mean time
for subjects on trials two and three of Section One with
trials two and three of Section Two were .88 for retarded
boys and .76 for normal boys. Both of these correlation
coefficients were significant at the .01 level of signifi
cance.
In his discussion of the analysis of covariance,
Lindquist (1953, p. 318) has indicated how the regression
coefficient derived from such correlated performances may be
Vised to adjust criterion measures for initial inequities in
63
|
performance. By means of the coefficient of regression of
the initial performance speed in Section Two on the initial
speed in Section One, it was possible to calculate and ad
just criterion measures on each trial of Section Two for
each subject. Similarly it was possible to adjust criterion
measures on each trial of Section One for each subject on
the basis of the coefficient of regression of the initial
performance speed in Section One on the initial performance
speed in Section Two. These adjustments in time scores were
made only for retarded groups. Time scores for all groups
were converted to reciprocals for the statistical analysis.
Hypothesis 1
Hypothesis 1 concerns only the performance in Sec
tion Two of the Turning Task. The obtained and adjusted
pneans on each trial of Section Two for the four groups is
presented in Table A in Appendix B. Figure 3 depicts the
mean performance speed of each of the four groups over each
of the twelve trials using the adjusted means for the re
tarded groups.
It should be remembered that each performance in
Section Two is preceded by a performance in Section One and
while every performance in Section Two is rewarded, trials
4, 6, 7, 9, 10, and 11 are preceded by nonreward on Section
One for the partial reinforcement groups. A 2 X 2 factorial
analysis of variance was done on the data of trial 3 of Sec
tion Two (Edwards, 1960). Trial 3, of course, occurs before
O Retardates
0 Normals
11.00
Continuous Reward
Partial Reward
8 i°-°° .
%
g
g 9.00 .
p .
c o
a
o
| 8.00
X)— o — o
7.00
6.00 ,
12
TRIALS
Figure 3. Performance Speed in Section Two of Turning Test
65
any experimental variation and thus it can be assumed that
[within intelligence groups partial and continuous groups are
similar. The analysis of these data is summarized in Table
B in Appendix B. The normal subjects performed signifi
cantly more rapidly than the retardates while, as expected,
the effect of reinforcement was not significant.
| A trend analysis of trial means on Section Two over
trials 4 through 12 was performed and the analysis of these
data is summarized in Table C in Appendix B. Normals per
formed significantly more rapidly than retardates and per
formance was significantly more rapid at the end of training
than at the beginning for all groups. While the reinforce
ment effect was not significant, the triple interaction of
intelligence X reinforcement X trials was significant at the
i
.10 level.
A rather striking difference in psychomotor function
between the retarded and normal boys of the same age is
observed in the above data; the retarded boys performing
consistently slower. While practice apparently tends to
improve the performance of both groups over the last nine
trials, the effect of nonreward immediately preceding a
trial did not appear to have a significant effect.
The curves in Figure 3 indicate that the "frus
trated" normal boys frequently performed slower than their
control group. The retarded boys seem to have responded as
predicted to Amsel's frustration theory and differently to
66
nonreward than the normals since the "frustrated" retarded
boys have mean speeds which are faster than the continuously
rewarded retarded boys on each trial on Section Two follow
ing the first nonreward experience. This tendency seems to
be reflected in the triple interaction effect mentioned
above.
The data for each intelligence group was analyzed
separately and the summaries of these analyses are reported
in Tables D and E in Appendix B. While the reinforcement
effect was not significant for either normals or retardates,
the F ratio for the interaction of reinforcement X trials is
considerably larger for the retarded than the normals and is
significant at the .10 level of significance.
Within the significance level established for the
idata analysis of the study there is no support for Hypothe
sis 1 although there is some suggestion that the retarded
boys may have responded differently to frustrative nonreward
than did the normal boys, and that the performance of the
retardates is in the direction predicted from Amsel's theory.
Hypothesis 2 >
Hypothesis 2 concerns only the performance in Sec
tion One of Task I. The obtained and adjusted mean of each
trial of Section One for the four groups is presented in
Table F in Appendix B. Figure 4 depicts the mean perform
ance speed of each of the four groups over each of the 12
trials using the adjusted means for the retarded groups. It
MEAN PERFORMANCE SPEED I N 1/SEC
11.00 .
10.00 •
9.00 .
8.00
7.00 .
6.00 ,
0 Retardates
A Normals
Continuous Reward
Partial Reward
-o
I ■ T ----- 1--------- 1------- I ------- I------- 1--------»— ' — ' ■ ■ > I | l
1 2 3 ^ 5 6 7 8 9 10 11 12
TRIALS
Figure Performance Speed in Section One of Turning Test o»
68
1
jwill be remembered that trials 4, 6, 7, 9, 10, and 11 were
j
not rewarded for the partially reinforced groups and that
every performance in Section One was followed by a perform
ance in Section Two of Task I.
Figure 4 alone suggests that there is no support for
Hypothesis 2. At least for the retarded groups it is appar
ent that, contrary to the prediction, the partially rewarded
group performed more rapidly in the later trials than did
the continuously reinforced group.
An analysis of variance of a 2 X 2 factorial nature
was done on the data of trial 3 of Section One. The analy
sis of these data is summarized in Table G. The main effect
of intelligence level is very significant with the perform-
l
ance of the normal subjects being significantly faster than
that of the retarded subjects. As expected, the effect of
reinforcement was not significant. It can be assumed that
•within intelligence groups, the partially and continuously
.rewarded groups were initially similar prior to experimental
manipulation of reward.
A trend analysis of trial means on Section One of
Task I was performed over trials 4 through 12, similar to
the analysis of the data on Section Two. This analysis is
summarized in Table H in Appendix B. The findings are very
similar to those found for the data on Section Two. The
normal subjects perform faster than the retarded subjects
and speed of performance increases with training. The
69
reinforcement is not significant although the triple inter-
I
'action of intelligence X reinforcement X trials falls just
slightly below the .05 level of significance. The data for
each intelligence group were analyzed separately and the
summaries of these analyses are reported in Tables I and J
in Appendix B. While the interaction of reinforcement X
[trials is not significant at the .05 level for either
normals or retardates, this F ratio is larger for the re
tarded group and falls just short of the significance level
established for the study. Contrary to prediction, the
Ispeed of "frustrated" retardates tended to increase more
than the "non-frustrated" retardates.
, Thus there is no evidence that periodic nonreward
I
tended to decrease speed of subsequent task performance and
consequently no support for Hypothesis 2. In fact, for
I
I
retarded boys these data indicate a nonsignificant tendency
in the opposite direction.
Hypothesis 3
!
The index of persistence was trials to extinction on
the Seguin Form Board and was the sole data for investigat
ing Hypothesis 3. Only the first four trials on this second
task were rewarded, nonreward beginning with the fifth trial
land continuing through trial 54 when the experimenter dis
continued the task if the subject had not stopped earlier.
Only 5 of the 56 subjects persisted for 50 nonreward trials
and four of these were subjects who had received partial
70
reinforcement in Task I. Since persistence has been inter
preted as continuing activity in the face of nonreward,
trial five, the first nonrewarded trial was not counted as a
trial to extinction.
Table 2 contains the mean number of trials to ex
tinction for each of the eight groups.
TABLE 2
MEAN NUMBER OF TRIALS TO EXTINCTION
ON SEGUIN FORM BOARD
Task Is Reinforcement Condition
Seguin
Response
Normal Retarded
1
Continuous Partial Continuous Partial
Switch 17.6 26.6 21.7 14.9
Button 13.7 26.6 10.6 19.1
There was a significant correlation between means
I
and variances in these data on trials to extinction and a
logarithmic transformation was necessary to analyze these
data by a parametric test of the analysis of variance. The
experimental design is suitable for a 2 X 2 X 2 factorial
model, intelligence (normal vs. retarded) X reinforcement
(partial vs. continuous in Task I) X completion response on
geguin (button vs. switch). __
71
j The two part analysis of variance of this factorial
I
design is summarized in Table K in Appendix B. The F ratio
for the over-all treatment effect falls at the .10 level of
significance. Although this is below the significance level
established for the data analysis of the study, it does not
'
necessarily indicate a terminal test with the factorial
{design and the subsequent partitioning of sums of squares
and significance tests can be justified (Edwards, 1960).
The major justification is based on the position that the
hypothesis to be tested does not predict differences between
all treatments. Furthermore, Hypothesis 3 predicts ordered
(differences among the major comparisons to be made. Page
(1963) has pointed to the power weakness of the F test in
t
the testing of ordered differences among means.
None of the main effects of intelligence, reinforce
ment history nor completion response to the Seguin were
i
i
bignificant and only the reinforcement history effect
approached significance. However, the interaction of type
of response to the Seguin and reinforcement condition on
jPask I was significant at the .05 level. An inspection of
the raw data in Table 2 does suggest that for both normal
and retarded subjects with a history of continuous rein
forcement in Task I, there appeared to be greater persist
ence when the switch was used in the seguin than when the
button was used. This is an unexpected finding which sug
gests either some intrinsic value for the switch response or
jthat some learned value accrued in Task I even among the
groups continuously rewarded on both button and switch in
that task. Such a tendency would obviously tend to counter
act the transfer effect of partial reinforcement from Task I
which only occurred for the button response.
The planned comparisons between various treatment
groups on the trials to extinction data were made with t
tests.^ For both the normal and retarded groups, greater
persistence on the Seguin is found in favor of the groups
partially rewarded on Task I only when the completion re
sponse to the Seguin was button pushing. At of 2.44
j
(df =48; p < .01 in a one tail test) was obtained in the
comparison of the two groups of normal boys with different
I
reinforcement histories when both used the button response
jSeguin the other two groups of normal boys with different
!
reinforcement histories did not show any difference in
trials to extinction. In this comparison a t of .79
i
(df =48; p > .05 in a one tail test) was obtained.
the square root of the error mean square in the analysis of
variance divided by the number of observations for each mean
multiplied by the sum of the coefficients of the treatment
means (Edwards, 1960).
bn the Seguin. However, when the switch was used on the
1 ,
d.
The t_ was computed using the formula; t = in
which d. is the difference between means and s,
i d
Parallel findings were obtained with the retarded
groups. A t of 1.84 (df = 48; p < .05 in a one tail test)
was obtained in the comparison of the two groups of retarded
boys with different reinforcement histories when both used
the button response on the Seguin. However, when the switch
I
was used in the Seguin, a t of 1.17 (df =48; p > .05 in a
one tail test) was obtained. This last difference while not
significant is opposite to the predicted direction.
Thus neither the normal nor the retarded groups who
received partial reward in Task I showed greater persistence
on the Seguin if the completion response on that task was
the switch. It will be remembered that it was only the
button response which was directly associated with partial
i
reinforcement in Task I— the switch was continuously rein-
jforced for all groups.
I
The data suggest support for Hypothesis 3 but with
I
some qualifications. It was predicted that persistence
would be greater on the Seguin for all partially reinforced
i
groups and that this persistence would be greater when the
button response was used rather than the switch. The data,
however, suggest that persistence is greater only when the
button was used. Thus the transfer of the partially rein
forced extinction effect seems to occur only when the two
tasks contain identical components in their completion
response and that there was little generalization of the
partial reinforcement effect.
j 74
I Hypothesis 4
i
As mentioned earlier, the partial reinforcement
t
group of retarded boys performed slower on Task I prior to
any experimental manipulation of reinforcement. This group
also performed slower than their control group on all trials
on the Seguin. Since the initial performance speed on
Task I was correlated with speed on the Seguin for the
jretarded group (r = .67; df = 26, p < .01) it was possible
to adjust the criterion measure of speed on the Seguin for
fhe retarded groups on the basis of the coefficient of re
gression on initial performance speed of the Seguin with the
initial performance on Section One of Task I. Such an
adjustment is valid only on the assumption that the differ
ences on the Seguin between the two groups of retarded boys
reflect some basic inequities in motor performance speed
father than treatment effects from Task I.
I
I
i Figure 5 depicts the mean performance speed of each
of the four groups for the first eight trials of the Seguin
jporm Board. Table L in Appendix B presents the mean time
j?er trial for each of the groups over these eight trials
including the obtained and adjusted means for the retardates.
It will be remembered that the first four trials are re
warded for all groups and all subsequent trials are non
rewarded.
Since Hypothesis 4 is concerned with the effect of
nonreward on speed of subsequent performance on the Seguin,
MEAN PERFORMANCE SPEED I N 1/SEC
23.00
Q Retardates
^ Normals
22*00
21.00 Continuous Reward History
Partial Reward History
20.00
19.00
18.00
17.00
16.00
15.00
14.00
13.00
12.00
11.00
4 6 8 1 2
3 5 7
TRIALS
Figure 5. Performance Speed on Seguin Form Board
" 76
jthe relevant performances are those on trials 5 and 6. The
jfifth trial is the first nonrewarded trial and thus a FE
would not occur until the sixth trial. A comparison between
speed of performance on trials 5 and 6 was made for each of
the two groups with different histories of reinforcement.
i
None of the four groups showed any significant change in
jperformance speed from trial 5 to 6. Thus there is no sup-
jport for Hypothesis 4.
Chronological Acre and Performance
Within the general developmental period of childhood
and adolescence, motor performance is correlated with age.
While the present study used subjects within a relatively
narrow age range, the relationship between age and perform
ance was investigated. The Seguin Form Board is actually
psed as a developmental task related to intellectual devel
opment although it loses its discriminatory powers about
i
jage ten years. As such a task, the index of performance is
the fastest speed the individual performs the Seguin task in
|
any of three trials. Using such an index with the present
i
data, the normal group of boys have a mean speed of 13.50
seconds on the Seguin while the retarded boys have a mean
speed of 16.53 seconds. However, the variance of the raw
scores for the retarded group is four times as large as the
variance for the normal groups. A reciprocal transformation
was used and a t of 5.07 was obtained (df = 54; p < .005)
between retarded and normal groups.
j For the normal group alone, the correlation coeffi
cient for the fastest of the first three scores on the
Seguin with age is -.21 (df = 26; p > .05). There is a
similar nonsignificant correlation of -.16 for the retarded
boys. There was no correlation between age and performance
I
on Task I for either normal or retarded groups,
j Zero order correlations for both groups were found
for the relationship between age and trials to extinction.
(Neither was there any correlation between speed of perform
ance on the Seguin and trials to extinction for either
(group. Table M in Appendix B contains the values of these
various coefficients. A general conclusion from these find
ings appears to be that chronological age was not related
i
to performance speeds nor to the persistence index. How-
l
jever, as reported in the earlier section, performance speed
on both tasks was related to intelligence.
I
Interview Analysis
A qualitative rather than a quantitative approach
|
was taken to the responses to the interview questions. One
|
bf the reasons for this was that a number of the questions
were open ended and produced a variety of responses diffi
cult to categorize. It will be remembered that the inter
view was abandoned in the pilot study with the institutional
retardated subjects because of their inability to comprehend
and respond. Even within the present study many of the re
tardates seemed to respond very arbitrarily to the questions
78
and could offer very little justification or elaboration of
i
[
their responses.
i
A tally of the responses indicated virtually no dif
ferences between the partially rewarded groups and the con
tinuously rewarded groups. On those questions in which sub
jects were to indicate one of two possible responses, it was
feasible to collapse the tallies into a fourfold table with
normal and retardates as the other dichotomy and use chi
square as described by Guilford (1956, p. 551).
Another reason for the qualitative approach to the
analysis of the interview responses was that a correct
I
interpretation of a subject’s response to one question could
only be determined from its relationship to answers to other
questions. For example, the first question, "Do you think
you maybe could have won more marbles?" was asked to deter-
i
mine if the subject interpreted the marble reward as depend
ent upon his performance or some extrinsic factor. A
response of "No" did not necessarily mean that the reward
factor was "rigged" in the subject's estimation since a
response to question three, "Why do you think you didn't
always get marbles?," might indicate that his original "No"
meant he was not capable of performing well enough to get
jnore marbles although the marble reward was still contingent
upon performance. Only two subjects, one retarded and one
normal but both in the continuous reinforcement groups,
indicated in a clear cut way that the marble reward was
79
i
|"rigged" in some way and was not contingent upon their per
formance. Three other subjects, all in the normal groups,
implied that external factors such as "the machine ran out
of marbles" might have played some role yet inconsistently
{indicated that they could have won more marbles or that
i
skill played some role in reward. One retarded subject was
junable to explain any of his answers.
The majority of subjects, 41 of the 56, indicated in
their verbal reports that reward was contingent upon their
skill. Only one subject indicated that the experimenter
himself was responsible for rewards, "Two reasons— 'cause I
messed up a couple of times and you didn't give me any."
Others who suggested the games were "rigged" blamed the
{apparatus.
i
! Of the 41 subjects who indicated that skill was
{responsible for reward, the majority, 27, indicated that
|
speed of performance was the critical factor. Other re
sponses indicated "too many mistakes" or some factor of dex
terity. It was interesting that four of the retarded but
pone of the normal subjects tended to blame themselves, "I
!
didn't work hard enough."
Twenty-seven of the 28 retarded subjects indicated
jtheir preference for Task I in answer to question 4, "Which
game did you like most to do?" However, 9 of the 28 normal
subjects indicated a preference for the Seguin Form Board.
All subjects had won more marbles on Task I and that task
80
lhad ended with reward while the Seguin ended with nonreward.
When these data were collapsed into a fourfold table, normal
vs. retardate and Seguin vs. Turning Test, significantly
more retardates than normals had a preference for Task 1
(the chi square is significant at the .05 level).
Question 5, "Which game did you like least to do?,"
was intended as a check on question 4. Only the retarded
groups responded with any inconsistency. Eight of these
subjects indicated the "least liked" task was the same as
the "most liked." It was not until about half of the sub
jects had been tested that it was realized that the retarded
boys frequently did not know what the word "least" meant.
A few of the subjects asked what the word meant, others
seemed to look puzzled at the question and then responded
iarbitrarily. On the other hand, there was little hesitancy
xn responding to question 5 among the normal subjects.
Question 6 was on the order of a "resumption of
task" question, "If you came back tomorrow which one game
Would you like to play?" Ten of the 28 normal subjects
jindicated the Seguin Form Board despite the fact that they
had received less reward on this task. Only two of the
retardates selected the Seguin. When the tallies were col
lapsed into a fourfold table, normal vs. retardate and
Seguin vs. Turning Test, the chi square is significant at
the .05 level.
81
The majority of both normal and retarded subjects
selected Task I as the game on which more marbles could be
won, question 7. Of the few normal subjects who indicated
they would select the Seguin to win more marbles, some logi
cal reasons were indicated such as it being the preferred
game or that a different approach would be used in subse-
jquent trials. The few retardates who selected the Seguin
I
tended to give arbitrary responses, "It's the second best
game."
Question 8 was concerned with the subject's verbal
report of persistence as it might relate to the use of the
button or switch as the completion response on the Seguin.
jFhe question is repeated here for convenience.
!
Do you think if you had used this button (or switch
depending on which had been concealed during the task
proper) you would have
i a. played the game more times. Why?
b. quit the game sooner. Why?
c. played the game the same number of times
as you did.
The question was directed at the ability of the sub
jects with a history of partial reinforcement to recognize
land verbalize that continuous reward was associated with the
switch and intermittent reward with the button and the
effect this would have upon persistence on the Seguin. This,
Of course, is related to the major hypothesis of transfer of
the partial reinforcement effect for which the performance
data offered some support. The verbal responses, however,
indicated that about the same number in each group regard-
less of the reinforcement histories said they would have
played the Seguin longer if allowed to use the completion
response that had not been available to them. Thus there
seemed to be no verbal acknowledgment that the button was
related to greater persistence. Only one subject indicated
the relationship of the button and switch on the Seguin to
[the button and switch on Task I. This was a normal subject
|who had partial reward on Task I and the button response on
the Seguin. In answering question 8 he indicated he would
have played longer with the switch, "I'd get more marbles
with the switch because that's what I was getting 'em off on
that" (points to Task I). This, of course, is directly
opposite to the prediction of the partial reinforcement
I
effect. What is interesting is that this was one of the
five subjects who persisted for 50 nonrewarded trials on the
iSeguin.
Question 9 was similar to question 8, but pertained
more directly to the value established for the button vs.
the switch response, "Which one the switch or the button
yrorks better or are they both the same?" No subject with a
partial reinforcement history selected the response he
actually used on the Seguin as better than the alternate and
either indicated the alternate worked better or both re
sponses worked equally well. However, a similar tendency
was seen in the answers of the groups with a history of con
tinuous reinforcement. It seems likely that the answer to
j 83
i
this question was influenced by the varying degree of non-
jreward on the Seguin alone without much influence from the
experiences on Task I. Reasons given for a preference of
either switch or button usually made reference to one being
"easier" than the other.
t
Responses to question 10 indicated no greater ten
dency for the partially reinforced groups to report they
performed slower (or faster) on Section One of Task I than
did the continuously reinforced groups.
Question 11 was asked only of the partially rein
forced subjects, "When you didn't get a marble here some
times what did it make you do?" Being an open ended ques
tion it elicited a variety of responses from "made me mad a
j
little bit" to "try harder." The retardates had more diffi
culty answering the question while the normal subjects more
frequently made reference to "working faster."
The last question was also asked only of the parti
ally reinforced groups.
i On this part, the second half (pointing to the second
segment of Task I), when you didn't get a marble here
! (pointing to the button response) do you think you
worked
a. faster. Why?
b. slower. Why?
c. the same as when you got a marble here.
All but one of the normal subjects reported that
nonreward made them work faster on Section Two. However,
the retardates' answers seem to be rather evenly distributed
over the alternatives.
CHAPTER VI
DISCUSSION
Frustration Effect
Hypotheses 1 and 4 were directly concerned with
Frustration Effect (FE) as it has been described and ex
plained by Amsel.
In Task I, the Turning Test, the continuously rein
forced subjects increased their speed of performing the task
over the twelve trials on Section Two in a similar manner as
did the partially reinforced subjects. This was true for
both retarded and normal groups and was interpreted as a
effect. The occurrence of an FE would have been
supported had the speed of performance of the partially
i
rewarded groups increased more than that of the continuously
reinforced groups following the introduction of nonreward at
the end of Section One. The data do suggest such a tendency
for the retarded subjects. The partially rewarded group of
retardates consistently performed at faster speeds following
the first nonreward experience than did the continuously
rewarded retarded group. On the other hand, among the
normal groups, the performance of the partially rewarded
subjects frequently but not consistently was slower than
that of the continuously rewarded group. However, the sta
tistical analysis only offered evidence of a practice effect
for all groups and a significant difference in performance
speed between normal and retarded groups.
Hypothesis 4 had predicted a greater FE demonstrated
by increased speed of performance on Task II, the Seguin
Form Board, following the first nonrewarded trial (trial 5)
for those subjects with a history of continuous reinforce
ment on Task I compared to those subjects with a history of
partial reinforcement on Task I. There was no significant
change in speed of performance following the first nonre
warded trial for any group although the speed of performance
pf all groups increased during the first three trials which
I
was probably attributable to the effect of practice at that
task.
Thus, on two separate tasks there was little evi
dence for Amsel's FE which has been so frequently found in
animal studies and studies with young children. There are a
I
humber of possible explanations for the failure to demon
strate the FE.
There might be a question of whether the frustration
occurred and its effect was dissipated too quickly to be
Reflected in performance speeds of the two tasks used in
I
this study. The FE is attributable to a momentary increase
in drive as a consequent on nonreward in previously rewarded
subjects and thus the time interval between frustrative
)
nonreward and subsequent task performance is a crucial
i f actor. The FE is typically found when the interval is
under one minute and usually in the range from 10 to 45 sec
onds. The intratrial interval in Task I and the intertrial
Interval in the Seguin were within this range, and it there-
i
fore seems unlikely that if frustration and heightened drive
jwere present that the drive would have diminished prior to
i
the next task performance.
The present study differed in two major respects
from other studies dealing with Amsel's theory at the human
level, because more complex instrumental tasks were per
formed for reward and because the present study used older
phildren as subjects. Task complexity may have been a
factor which masked the reflection of FE in terms of time
I
scores. The old Yerkes-Dodson law (Brown, 1961) has found
0
I
I
some additional support in recent literature on the disrup
tive effects of heightened drive (anxiety) on complex but
jnot simple tasks. The Steigman and Stevenson (1960) study
i
Mentioned earlier suggested a possible decremental effect of
jnonreward on subsequent complex but not simple learning
i
1
tasks. However, the results of the present study do not
suggest any deterioration of performance as a result of non-
jreward and on Task I. On the contrary, there is even a non
significant tendency for the partially rewarded retarded
group to perform at faster speeds than the continuously
i 87
rewarded retarded group following the first experience of
Inonreward.
While neither the Minnesota Rate of Manipulation
Test nor the Seguin Form Board are particularly difficult
tasks they are considerably more complex than the simple
I
lever pulling tasks which the majority of studies have used
jin investigating FE with children. In fact, the instruc
tions for the administration of the Seguin (Arthur, 1930)
caution against too great an emphasis upon speed as this may
be detrimental to performance. One possible explanation for
the failure to obtain the predicted increase in performance
speed following nonreward may have been due to heightened
drive resulting in some decremental effect on dexterity.
i
Another and possibly interacting factor may have
been that older subjects having greater capacity for sym-
!
bolic processes can respond in a greater variety of ways to
^onreward in these more complex tasks. There are only a
limited number of hypotheses that an individual can enter
tain about his effectiveness in getting a lever to deliver a
Reward. He can increase or decrease speed or increase or
decrease vigor. However, in such a task as the Seguin one
can entertain many hypotheses concerning the contingency of
feward on performance. While the majority of subjects did
j
give verbal reports that speed was important, they also
indicated that making errors, or that dexterity and sequence
of block placement might play a role, as well as all the
88
factors that could vary with button pushing and switch pull-
j
|
jing. Such an explanation steps out of the Amselian frame
work and more into the area of cognitive learning theory.
Yet it does seem possible that heightened drive due to frus
tration in complex tasks might be utilized in increased cog-
i
nitive functioning and variation in task solution rather
than a straightforward increment in performance. The
younger or retarded child with more limited symbolic proc
esses might be more likely to demonstrate the more straight
forward effect of nonreward. As mentioned, in the present
j
study the retarded group did perform in the predicted direc
tion following nonreward on Task I while the performance of
the normal group following nonreward on this task was fre
quently inferior to the continuously rewarded normal group.
Animal studies which have investigated frustrative
j
ponreward usually do not introduce nonreward until asymtote
!
level or performance has been reached. While the Penny
Study (1960) mentioned earlier, found evidence for the FE in
i
Children's lever pulling only after considerable training,
penny's group also showed evidence of fatigue. With massed
trials of a more complex task a decrement due to fatigue is
likely to be a confounding factor. In the present study,
ponreward occurred while performance was still improving
i
with practice, at least, in Task I. Thus there is another
point of divergence from the usual experimental design of
frustration studies.
89
! The findings of the present study concerning FE are
in no way interpreted to be counterevidence for the phenome
non as it has been described in the literature. The study
was designed to explore the possible occurrence of the phe
nomenon with older children and more complex tasks than have
been previously employed. The failure to produce the FE may,
■however, suggest caution in the generalization of Amsel's
I
frustration theory to more typical situations of nonreward
for the older child.
Anticipatory Frustration Responses
It will be noted that it is possible to describe
Amsel's FE without any reference to anticipatory frustration
responses which play such a significant part in his theory
as a whole. However, anticipatory responses are crucial to
I
jfrustration theory's explanation of the partial reinforce
ment effects. Hypothesis 2 of the present study concerned
the effect of such responses upon performance in Section One
of Task I.
t
Actually, Amsel (1958) has little to say about per-
I
I
formance speeds in runway 1 of his two-stage runway appar
atus. He is primarily concerned with running speed in run
way 2 following nonreward which he calls the frustration
motivated response. He refers to responses in the first
runway as the frustrated response and says only that in
early trials the anticipation of frustration should lead to
"increased variability and decreased vigor." It follows
90
that if anticipatory frustration responses develop in the
i
first section, then performance should show some decrement
at least until the time these responses become conditioned
to anticipatory reward responses. It is by this mechanism
that Amsel's theory can explain the empirical findings con
cerning poorer initial acquisition scores with partial
jrather than continuous reinforcement.
i
The Holder et al. study (1957) mentioned earlier
suggested that frustration due to retention in the midsec
tion of a two-stage runway resulted in slower running speeds
for rats in the predelay box section of the runway but
faster running in the postdelay box section. The authors
indicate the results are predictable from Amsel's theory in
that increased speed in the postdelay section is due to
(increased drive (FE) while the decreased speed in the pre
delay section could be due to anticipatory frustration.
Amsel does mention that partial reinforcement during
acquisition can lead to slower running speeds in the single
jrunway task. However, he suggests that such partial rein
forcement should have a diphasic effect in the sense that
running in early trials would be slower for the partially
reinforced animals due to anticipatory frustration responses
but in later trials these frustration responses would become
I
conditioned to anticipatory reward responses and the animals
would run fast or faster than the continuously rewarded
animals.
I 91
I
I Hypothesis 2 predicted only the initial decrement in
performance due to anticipatory frustration in Section One,
and predicted that performance of the partially rewarded
subjects would tend to be inferior to that of the continu
ously rewarded subjects. The data offered no support for
this hypothesis and tended to suggest the opposite tendency
;for the retarded group.
Animal studies such as the Holder et al. and many
of Amsel's studies use rather long intertrial intervals,
frequently as long as 24 hours between trials. There are
too many practical problems to allow the use of such long
intertrial intervals when human subjects are used. Ryan
(1965) has indicated that in lever pulling experiments with
i
children partial reinforcement may interfere initially with
i
some aspect of the response such as starting time but not
necessarily movement time. However, in the present study
the intertrial interval was small and it is possible that
heightened drive due to nonreward may still have influenced
Successive performances and masked any aversive tendencies
I
(toward the potentially frustrating button.
i
Transfer of the PRE
The data analysis did offer some support for trans-
i
fer of the partial reinforcement extinction effect in terms
of greater persistence on Task II following partial rein
forcement on Task I than continuous reinforcement on Task I.
However, the effect seemed to be limited and only occurred
92
when there were identical components common to both tasks.
In Task I it was only the button response that had
any frustration response associated with it and then only
for the partially rewarded groups. The switch response was
Always associated with reward for all groups. It was be
cause the button was associated with both anticipation of
frustration and anticipation of reward for the partially
rewarded groups that it was predicted these groups would
persist longer when the button response was used to signal
completion of the Seguin. The prediction was supported for
both the normal and retarded subjects.
The prediction that the partially rewarded subjects
in Task I would also persist long on Task II when the switch
I
was used to signal completion of that task was based on the
assumption that anticipatory frustration associated with the
^utton on Task I might also generalize to the switch. The
j
data did not support this prediction although there was a
tendency in this direction among the normal subjects. In
(fact, of the four normal subjects who persisted for all 50
jionrewarded trials on the Seguin, all had a history of par
tial reinforcement in Task I, two having used the switch and
two having used the button as the completion response on the
jseguin.
i
Among both the normal and retarded subjects who re
ceived continuous reinforcement on Task I, there was a non
significant tendency for these groups to persist longer on
93
the Seguin when the completion response to that task was the
i
switch rather than the button. This was an unexpected find
ing and if indeed, the switch did have some greater attrac
tiveness on Task II, it would have tended to reduce the
•transfer effect of partial reinforcement for the button
I
response to Task II.
j The above interpretation is a plausible explanation
for the significance of the interaction effect of type of
response X reinforcement in the analysis of the data con
cerning trials to extinction on Task II and the failure of
the main effect of reinforcement history to reach the sig
nificance level established for the study.
Since the study did not provide controls for this
i
eventuality by balancing the position of switch and button
in Task I, only post hoc explanations can be offered. The
switch may have had greater attractiveness because of its
placement, side front compared to the top front placement of
the button. Also the switch was of the variety more com
monly used in the electric light switches in the home than
Was the button. In the interview responses, as many sub-
!
jects said the button worked as well as the switch, however,
by the time the interview as administered the subjects had
had unequal experiences of reward and nonreward with the two
i
responses.
There is a possibility that the tendency among the
continuously rewarded groups in Task I to persist longer on
94
I
the Seguin when using the switch rather than the button
could have been related to the fact that the switch came at
the completion of a full trial, that is, both sections of
Task I. The success of a single marble reward at the end of
Section Two of Task I which was always associated with the
switch may have had greater value to the subjects than the
garble reward at the midsection that was always associated
i
(with the button. It might also follow that nonreward at the
midpoint of the trial would be less frustrating than nonre
ward at the end of the final section. The experimental
(design did not call for any manipulation of reward at the
end of Section Two and there are no data in the present
study to investigate such a possibility. However, it is a
I
possible explanation which perhaps should concern other
[investigators attempting to use parallels of Amsel's two-
jstage tasks with human subjects. For the rat, food reward
at the mid-goal box may have the same value as food at the
final goal box in the two-stage runway apparatus. For the
I
human subject successful completion of the final half of a
jtask may be of greater value than successful completion of
I
the first half even though the tangible reward given for
each half is the same.
Persistence or Perseveration
While resistance to extinction of a response has
been associated with the term persistence (Feather, 1962), a
term which has a socially desirable connotation, it might
95
just as logically be associated with the term perseveration
which has an undesirable connotation. The continuous repe
tition of an activity in the face of nonreward connoted by
the word perseveration suggests a rigidity in the adaptation
jthat one might expect in the less intelligent subject. In
the present study, the data suggest that the normal subjects
jshowed greater resistance to extinction. The use of such
terms could easily invite sophistry with the term persevera
tion being used when retarded subjects show greater resist-
ance~to extinction and the term persistence used when more
intelligent subjects show the same behavior. Verbal reports
i
may suggest the nature of whatever cognitive variables are
involved in resistance to extinction in human subjects.
!
In the present study the great majority of both
I
formal and retarded subjects indicated in their answers to
I
jthe interview questions that more marbles could have been
won. However, twice as many retarded as normal subjects
indicated that it would not have been possible to win more
inarbles and this could account for the slight tendency of
jthe retarded subjects to show less resistance to extinction.
The fact that the majority of subjects did indicate even
after they had quit the task that further rewards were still
jceasible tends to remove the connotation of the purposeless,
I
stereotyped behavior during extinction perhaps suggested by
the term perseveration.
96
Frustration Tolerance
As mentioned earlier, Amsel has made a brief comment
that his concept of anticipatory frustration and its rela
tionship to resistance to extinction might be the learning
theory counterpart of the concept of frustration tolerance.
Although Amsel does not elaborate this point, the comment
suggests that intermittent frustration due to nonreward on
one task could be conditioned to approach behavior and would
strengthen approach behavior in the face of the frustration
of nonreward on a subsequent activity. The explanation fits
rather nicely the observation that the overindulged,
"spoiled" child is most frequently the least tolerant of any
frustration of his needs. If one were to assume that the
j
retarded child has more experience with frustration due to
i
jnonreward, at least in school activities, then the predic
tion might be that the retarded child would show greater
l
persistence. Yet observations of the retarded child's be
havior in the classroom or testing situation, and in the
experimental situation used for the present study seem to be
(quite contrary to this prediction. Amsel's model, however,
!
demands that nonreward must be interspersed with reward in
order for aversive tendencies to be conditioned to approach
tendencies.
i
The data did offer some support for Amsel's model of
frustration tolerance but only when the two tasks involved
had similar components. Only then did "frustration toler-
1 97
I
jance" established in the first task transfer to the second
task.
Motor Performance
One of the consistent findings of this study was the
inferior performance speeds of the retardates on both tasks.
While the literature in this area (Malpass, 1963) indicates
i
‘ that moderately and severely retarded individuals do show
less proficiency in motor skills than normals of the same
age and sex, the correlation between IQ and motor skills is
generally very low when mildly retarded, noninstitutional
retardates are the subjects. Malpass (1960) states that
some investigators have hypothesized that an organic condi
tion might be a responsible factor common to the mental and
motor deficiency of retardates. One might expect this in
i
jthe severely retarded since neuropathology is more fre
quently established in the low grade versus the high grade
defective (Malamud, 1964).
In the present study, however, with the exception of
t
one subject, the IQ range in the retarded group was from 60
ito 75. These subjects were matched in chronological age
within six months with normals and 75 per cent of the retard
ates were matched with the normals within three months.
Both Malpass (1960) and Francis and Rarick (1959)
have shown low order positive correlations between IQ and
various motor skills for boys in EMR classes of an age range
from 8 to 14jyears. Francis and Rarick used rather gross
98
motor performance tests such as grip strength, running,
jumping and balancing. They reported no greater intercorre
lations among the tasks for the retardates than for the
normal subjects. However, the study did find that on almost
all measures the EMR children were between two and four
years behind the published age norms for normal children of
jthe same chronological age.
i
Malpass used the Lincoln-Oseretsky Motor Development
Scale in comparing three groups of children in the same age
range as in the Francis and Rarick study. The performances
of institutionalized mental defectives were compared with
EMR children and children with normal IQ scores. The two
retarded groups did not differ from each other but both were
i
inferior to the normal group. Malpass also found that motor
skills frequently correlated with IQ scores within the re-
i
tarded groups but not with the normal group.
A study by Cantor and Stacey (1951) is one of the
few investigations that has been concerned with manual dex
terity in the retardate. The study compared the performance
pf institutionalized male retardates, ages 14 to 18 years,
i
on the Perdue Pegboard with the performances of noninstitu
tionalized normals. They report significant differences in
jscores even when the comparison is only between the normals
&nd the retardates in the 70 to 82 IQ range. However, the
study is open to many criticisms. The normals were adult
men, either veterans or men in industry. Furthermore, there
Jis no explanation why boys with borderline IQ scores were in
an institutional setting.
Dawson and Edwards (1965) have recently shown that
when retardates are matched for height and weight with
normals rather than chronological age, the difference be
tween the groups in grip strength virtually disappears.
iThey conclude that "physiological development may account
i
for a significant portion of the variation in strength for
normal and retarded children" (p. 223) and suggest this may
also be responsible for other reported differences in motor
skills between normals and retardates.
In the present study a significant correlation be
tween performance speeds on the two tasks was found for the
!
retarded group but not for the normal group which is con
sistent with the findings of Malpass. The variances of time
i
i
scores are frequently greater for the retarded groups than
■for the normals, a fact which creates certain problems for
statistical comparisons of normal and retardates, but per
haps also a fact which may account for the larger correla
tion coefficients reported between performance and IQ scores
for retardates than for normals.
The fact that the retarded did take longer to per
form each task in the present study could well account for
their tendency to show less persistence on the second task.
, 100
I
I
J The Interview and Behavioral
| Observations
Throughout the discussion of the quantitative find
ings, occasional references have been made to the responses
to the interview data. This interview material was used
cautiously. The responses of the retardates did frequently
seem to be arbitrary and inconsistent. "Don't know" was a
j
Ifrequent response given by the retardate to a question or
request to explain his answer.
One of the reasons for the interview was to deter
mine whether the subjects felt that the reward was contin
gent upon their performance or some external factor.
According to the verbal responses, only 5 of the 56 subjects
i
indicated that the reward was influenced by external factors
jsuch as the apparatus or the experimenter. Only two sub
jects stated this in unequivocal terms. However, it might
be somewhat specious to conclude that the remaining 51 sub-
I
jjects believed that reward or nonreward was always a conse-
l
guent of their performance. There might well be consider-
!
able reticence on the part of a 12-year-old boy in a school
i
I
Betting to accuse an adult he has never met before of "rig
ging" the games.
The placement of the interview at the conclusion of
fcoth tasks was deliberate so as not to influence the per
formance on either task. Yet performance oh the second task
introduced varying degrees of nonreward for the subjects
Bince they could continue the task as long as they wished.
101
This certainly could influence attitude and responses to the
'questions. Finally, the questions themselves had to be
somewhat veiled in order not to reveal the role of the
experimenter in controlling rewards, and through the com-
taunication of the subjects with each other at school bias
I
the performance of the subjects yet to be tested.
Actually none of the subjects seemed very concerned
about winning the prize for the most marbles. Only a few
asked what was the largest number of marbles won and no sub
ject made a serious attempt to count the marbles he had won
after the second task. Only one subject indicated he ex
pected to keep the marbles he won. Despite this the sub
jects generally appeared to be concerned about getting a
marble as a measure of success.
Nonreward during Task I frequently brought some
I
jexclamation such as, "Hey! I didn't get one," or "I didn't
!
do so well." After a nonreward experience one subject re
marked in consolation, "Well, it's one way of getting out of
Ischool." Yet that very subject persisted for only six non-
rewarded trials on the second task. Normal subjects tended
to make verbal comments after nonreward while the retarded
subjects tended to finger the apparatus or the marble tray,
powever, both normal and retarded subjects seemed to be less
talkative during the massed nonreward trials on the Seguin.
Smiles disappeared, and for the most part there seemed to be
a silent determination that ended in an abrupt, "OK, I quit."
102
! One of the normal subjects displayed quite interest-
i
ing behavior to nonreward and the tasks in general. The
anecdotal comments made by the experimenter immediately
after testing the boy read as follows:
Boy with glasses— eager to do tasks— seems a little
immature— always blames self for not getting marbles—
actually increased effort seems to make him clumsy—
despite all this he remains good natured— counts
marbles at the end of the Turning Test but not at end
of Seguin.
After each nonreward experience on Task I, the boy remarked,
"Oh rats!" On the third nonreward trial he said, "I don't
do so good on this side," a few trials later, "I've bad luck
on this side," and after the last nonreward trial, "nothing
in the last six times, it's broken." Although he persisted
i
[for 32 nonreward trials on the Seguin he would remark be
tween these trials, "I'm mad at myself" and later "I'm mad."
I
At the end of the 32nd nonrewarded trial on the Seguin he
remarked, "OK, we can stop now." The sequence of remarks
suggests that on each task the subject initially tended to
fclame himself for nonreward but gradually shifted the blame
!
{:o the apparatus or the experimenter.
Another subject also in the normal group did not
talk as much,yet after each nonreward experience he would
slap his side in exasperation and this increased in intens
ity with each succeeding nonreward trial.
These are anecdotal reports of the two most dramatic
examples of behavior which could be interpreted as frustra
tion in its more clinical_cqnnotatIon. That subjects do
103
respond emotionally to failure is hardly a unique finding.
I
;The observations, however, did suggest that nonreward fol
lowing prior reward was frustrating. The observations per
haps also lend some support for the explanation that if
frustrative nonreward did result in an increase in drive,
the augmented drive may have energized responses other than
those involved in the subsequent task demands.
CHAPTER VII
SUMMARY
The study being reported investigated the predictive
jpower of Amsel's theory of frustrative nonreward with chil
dren of both normal and retarded intellectual status. There
is now a considerable amount of support for Amsel's theory
from research at the infrahuman level and Amsel has sug
gested that his theory might offer an S-R learning theory
explanation of human behavior identified in general person
ality theory as persistence or "frustration tolerance."
Amsel’s theory defines frustration as an internal
state which is elicited by nonreward of a response which has
I
previously been regularly rewarded. In such a setting, the
theory predicts that nonreward will result in (1) an in
crease in drive level which energizes ongoing behavior and
I
is called the frustration effect (FE), (2) a stimulus state
pf frustration (Sp) and, in later trials (3) the development
of anticipatory frustration responses and their consequent
stimuli ( rf-sf)•
The present study was designed to investigate the
role of nonreward in the test performance of normal and
mildly retarded boys on two tasks which were adaptations of
_______ 104___
105
standard psychomotor tests. The study tested hypotheses
'concerning FE and the transfer of the partial reinforcement
extinction effect (PRE) from the first to the second task.
Amsel and previous investigators studying the FE
have used a two-stage runway apparatus with the dependent
variable being the running speed of rats in the second run
way following frustrative nonreward at the end of the first
;runway. The theory predicts that running speed should in
crease on the second runway following such nonreward and the
research findings have offered support for this prediction.
;In the research with young children investigators have fre
quently used a design of two successive lever-pulling tasks
as analogous to the double runway of the animal studies. In
the first part of the present study the Turning Test of the
Minnesota Rate of Manipulation Test was modified to the
extent it consisted of two successive and equal segments.
On completion of Segment One the subject was to push a but
ton for a marble reward and at the end of Segment Two he was
to pull a switch for a marble reward. This was Task I and
provided a test for the following two hypotheses:
1. Performance on the second half of a task will be
more rapid after intermittent frustrative nonreward at the
midpoint than after continuous reward at the midpoint.
2. Performance on the first half of a task will be
slower following intermittent frustrative nonreward at the
106
jmidpoint of the task than after continuous reward at the
midpoint.
I
Hull's learning theory states that continuous reward
of a response will lead to the development of anticipatory
^oal responses (r -s ) and as an extention of Hull's theory,
1
Amsel states that intermittent nonreward will lead to the
development of anticipatory frustration responses ( r£-sf)•
jwhen nonreward is introduced after some period of continuous
reward these two types of anticipatory responses should be
in conflict and result in "increased variability and de
creased vigor." This is the basis of the second hypothesis
of this study.
However, in later trials, the avoidance tendencies
!
bf the rf-S£ should be conditioned to the approach tenden-
I
jcies of rg”sg anc^ as a consequence of such partial rein
forcement during training there would be greater resistance
to extinction. This is Amsel's explanation of persistence
or the PRE. Amsel has suggested that through the mechanism
pf these anticipatory responses there might be some transfer
across tasks of the PRE and there is some support for this
i
notion in research done at the animal level.
Part two of the present study was concerned with the
transfer effect of partial reinforcement on the first task
to the second task. Task II was an adaptation of the Seguin
Form Board. Subjects were to signal their completion of
each trial on the Seguin by either pushing a button
107
identical to the one at the midpoint of the Turning Test or
Iby pulling a switch identical to the one at the final seg
ment of the Turning Test. Part two provided a test for the
two remaining hypotheses of the study.
3. Intermittent frustrative nonreward on a first
imotor task will lead to greater resistance to extinction
j(persistence in the face of continuous nonreward) on a sec-
pnd motor task than will continuous reward on the first
motor task.
4. Frustrative nonreward on a second motor task
will lead to greater increase in performance speed of imme
diately subsequent performance on this task if the experi
ence on the first motor task has been of continuous reward
rather than intermittent frustrative nonreward.
The subjects consisted of 56 Caucasian boys; 28 were
enrolled in regular classes in the public school and 28 were
enrolled in classes for the educable mentally retarded. All
subjects were between the ages 10 years 2 months and 14
years 3 months. While no specific hypotheses were made con
cerning the difference between these normal and retarded
boys and the study might be considered as having been done
with normals and replicated with retardates, the experi
mental design did offer some opportunity to investigate the
general assumption that retardates are less tolerant of
frustration and less persistent in—the face of failure.
108
!
| All subjects were instructed that a prize would be
i
given to the boy who won the most marbles. Twelve trials
were given on Task I, the Turning Test. A trial consisted
of completion of both segments. Half of the normal boys and
half of the retarded boys were assigned to 50 per cent re
ward treatment at the end of Segment One (button response),
jthe remaining subjects received continuous reward at the end
pf Segment One. All subjects received continuous reward at
the end of Segment Two (switch response) of the Turning
Test. The time interval between first and second segments
was approximately ten seconds.
Approximately three minutes after the twelve trials
on Task I, the Seguin Form Board was presented. One-half
of the partially reinforced groups on Task I were instructed
to signal their completion of the Seguin by pushing a but
ton, the remaining subjects signaled their completion of the
Seguin by pulling a switch. All subjects were instructed to
perform the Seguin as many times as they wished and to tell
the experimenter when they wished to quit the game. The
jreward conditions on the Seguin were the same for all
groups, the first four trials were rewarded, all subsequent
trials were nonrewarded. If the subject had not quit the
task by che 50th nonrewarded trial, the experimenter discon-
I
tinued the task. A brief interview of 12 questions was
administered to all subjects after the completion of both
tasks.
109
]
j The results did not offer statistically significant
i
i
evidence of a FE following nonreward on either the Turning
Test or the Seguin Form Board for any of the groups and
neither Hypotheses 1 nor 4 were supported. All groups,
retardates and normals, partially rewarded and continuously
rewarded showed evidence of a practice effect on both tasks
iand there was a significant difference in performance speed
bn both tasks between normal and retarded groups.
The data did show a nonsignificant tendency for the
partially rewarded retarded subjects to perform faster, as
predicted by Amsel's theory, on the Turning Test following
the introduction of nonreward than the continuously rewarded
retarded subjects. However, among the normal subjects,
j
there was some tendency for the partially rewarded subjects
to perform slower following the introduction of nonreward
i
;than the continuously rewarded subjects.
The introduction of nonreward at the completion of
the first segment of the Turning Test did not result in
{decreased speed of performance on subsequent performances of
jthis segment and, contrary to the prediction of Hypothesis
2, the partially rewarded retarded group tended to perform
jnore rapidly in later trials than did the continuously rein
forced retarded group.
i
The failure to find performance speeds on the two
tasks influenced by nonreward as predicted from Amsel's
theory was discussed in terms of differences between the
110
present study's experimental design and the previous studies
i
jwith younger children and simple tasks which have results
consistent with predictions from Amsel's theory. The re
tarded subjects did seem to have a more straightforward
ireaction to nonreward and one which was more consistent with
i
l
frustration theory. The results did suggest some caution in
jthe generalization of Amsel's theory concerning FE and the
Speed of task performance by older children. There was also
some evidence that the retarded subjects reacted differently
than the normal subjects to nonreward and more like the
i
younger children who have been subjects in previous investi
gations.
The results of the present study did offer some sup
port for Hypothesis 3. Transfer of the partial reinforce-
i
jment extinction effect across tasks was found for both the
j
Retarded and the normal subjects. Subjects who were par-
j
tially reinforced on the Turning Test persisted longer in
performing the Seguin Form Board but only when there were
I
identical components common to the response of both tasks.
jrhe partially reinforced subjects in Task I persisted
j
longer on Task II only when the button response was part of
the completion response on the second task. Normal subjects
showed a nonsignificant tendency towards greater generaliza
tion in the transfer across task of the partial reinforce
ment extinction effect. The results were discussed with
reference to the concept of "frustration tolerance."
t
REFERENCES
J.11
REFERENCES
Amsel, A. The role of frustrative nonreward in noncontin-
1 uous reward situations. Psychological Bulletin,
1958, 55, 102-119.
Amsel, A. Frustrative nonreward in partial reinforcement
I and discrimination learning. Psychological Review,
! 1962, 69, 306-328.
!
Amsel, A., and Roussel, J. Motivational properties of
frustration: I. Effect of a running response of the
addition of frustration to the motivational complex.
Journal of Experimental Psychology. 1952, 43. 363-
368.
Amsel, A., and Ward, J. S. Motivational properties of
frustration: II. Frustration drive stimulus and
frustration reduction in selective learning.
Journal of Experimental Psychology. 1954, 48. 37-47.
!
Amsel, A., and Ward, J. S. Frustration and persistence:
Resistance to discrimination following prior experi-
i ence with the discriminanda. Psychological Mono-
| graphs: General & Applied. 1965, J79 (4, Whole No.
597).
Arthur, G. A. A point scale of performance tests. Vol. I:
Clinical manual. New York: Commonwealth Fund, 1930.
Badia, P. Effects of drive, reinforcement schedule and
j change of schedule on performance. Journal of
j Experimental Psychology, 1965, 69, 292-297.
kerlyne, D. E. Emotional aspects of learning. In P. R.
Farmsworth, 0. McNemar and Q. McNemar (eds.), Annual
review of psychology XV. Palo Alto: Annual Reviews,
Inc., 1964. Pp. 115-142.
Bernstein, B. Extinction as a function of frustration drive
and frustration drive stimulus. Journal of Experi
mental Psychology, 1957, 54, 89-95.
113
Britt, S. H , and Janus, S. Q. Criteria of frustration.
Psychological Review. 1940, 47. 451-470.
Brown, J. S. The motivation of behavior. New York: McGraw-
Hill Book Co., 1961.
Brown, J. S., and Farber, I. E. Emotions conceptualized as
intervening variables— with suggestions toward a
theory of frustration. Psychological Bulletin.
1951, 48, 465-495.
Brown, R. T., and Logan, F. A. Generalized partial rein
forcement effect. Journal of Comparative & Physio
logical Psychology. 1965, 60. 64-69.
Bruning, J. L. Effects of magnitude of reward and percent
age of reinforcement on a lever movement response.
Child Development, 1964, 35. 281-285.
Cantor, G. N. Hull-Spence behavior theory and mental
deficiency. In N. Ellis (ed.), Handbook of mental
deficiency. New York: McGraw-Hill Book Company,
Inc., 1963. Pp. 92-133.
jCantor, G. N., and Ryan, T. J. Response speed related to
! reinforcement schedule: Report of errata and addi
tional analyses. Child Development. 1964, 3 5 . , 999-
| 1002.
jCantor, G. N., and Stacey, C. M. Manipulative dexterity in
mental defectives. American Journal of Mental
Deficiency, 1951, 56, 401-410.
(Child, I. L., and Waterhouse, I. K. Frustration and the
i quality of performance: I. A critique of the
Barker, Dembo, and Lewin experiment. Psychological
Review, 1952, 5J9, 351-362.
pavenport, J., and Thompson C. The Amsel frustration effect
i in monkeys. Psychonomic Science, 1965, 3 . , 481-482.
Dawson, W. W., and Edwards, R. W. Motor development ot
retarded children. Perceptual & Motor Skills. 1965,
21, 223-226.
Penny, M. R. The role of secondary reinforcement in a par
tial reinforcement learning situation. Journal of
Experimental Psychology, 1946, 3 6 . ' 373-389.
Dollard, J., Doob, L. W., Miller, N. E., Mowrer, D. H., and
Sears, R. R. Frustration and aggression. New Haven:
Yale University Press, 1939.
114
Dyal, J. Latent extinction as a function of partial rein
forcement of the running response. Psychonomic
Science. 1965, 3 . , 207-208.
Edwards, A. L. Experimental design in psychological
research. (Revised edition.) New York: Rinehart &
Company, Inc., 1960.
Elmes, D. The role of frustration in the extinction of a
running response. Psychonomic Science. 1964, 1 . ,
345-346.
jFeather, N. T. The study of persistence. Psychological
Bulletin, 1962, 59, 94-115.
Festinger, L. The psychological effects of insufficient
rewards. American Psychologist, 1961, 16_, 1-11.
Francis, R. J., and Rarick, G. L. Motor characteristics of
the mentally retarded. American Journal of Mental
Deficiency, 1953, 63, 792-811.
i
Greenbaum, M., and Levy, N. Effect of multiple stress and
thorazine on Amsel's frustration effect. Psycho
logical Reports, 1965, 1 6 . , 947-948.
j
iGrosslight, J. H., and Child, I. Persistence as a function
of previous experience of failure followed by suc
cess. American Journal of Psychology, 1947, 60,
j 378-387.
Guilford, J. P. Fundamental statistics in psychology and
! education. New York: McGraw-Hill Book Co., 1956.
Haner, C. F., and Brown, P. A. Clarification of the insti
gation to action concept in the frustration-
aggression hypothesis. Journal of Abnormal & Social
Psychology, 1955, 5 1 , , 204-206.
|
Heber, R. Personality. In H. A. Stevens and R. Heber
(eds.), Mental retardation. Chicago: The University
of Chicago Press, 1964. Pp. 143-174.
polder, W. B., Marx, M. H., Holder, E. E., and Collier, G.
Response strength as a function of delay of reward
in a runway. Journal of Experimental Psychology,
1957, 53, 316-323.
Holton, R. Amplitude of an instrumental response following
the cessation of reward. Child Development, 1961,
3 2 . , 107-116.
115
Hull, C. L. Principles of behavior. New York: Appleton-
Century-Crofts, 1943.
i
Hull, C. L. A behavior system. New Haven: Yale University
Press, 1952.
Hulse, S. H., Jr., and Stanley, W. C. Extinction by omis
sion of food as related to partial and secondary
reinforcement. Journal of Experimental Psychology,
j 1956, 52, 221-227.
Humphreys, L. G. The effect of random alternation of rein-
S forcement on the acquisition and extinction of con
ditioned eyelid reactions. Journal of Experimental
Psychology, 1939, 2 5 . , 141-158.
Humphreys, L. G. Acquisition and extinction of verbal
expectations in a situation analogous to condition
ing. Journal of Experimental Psychology, 1939, 25,
294-301.
James, W. H., and Rotter, J. B. Partial and 100% reinforce
ment under chance and skill conditions. Journal of
Experimental Psychology, 1958, 55, 397-403.
i
Jenkins, H. M. Resistance to extinction when partial rein
forcement is followed by regular reinforcement.
Journal of Experimental Psychology, 1962, 64, 441-
' 450.
Jenkins, W. 0 , and Stanley, J. C., Jr. Partial reinforce
ment: A review and critique. Psychological Bulletin.
1950, £7, 193-234.
Kirkpatrick, D. R., Pavlik, W. B., and Reynolds, W. F.
Partial reinforcement extinction effect as a func
tion of goal box size. Journal of Experimental
Psychology, 1964, 68, 515-516.
I
Lawrence, D. H., and Festinger, L. Deterrents and rein
forcements. Stanford, Calif.: Stanford University
Press, 1962.
jLawson, R., and Marx, M. Frustration: Theory and experi
ment. Genetic Psychology Monographs. 1958, 5 J 7 . , 363-
464.
Lewis, D. Partial reinforcement: A selective review of the
literature since 1950. Psychological Bulletin.
1960, 5 7 . , 1-28.
116
Lewis, D., and Duncan, C. P. Expectation and resistance to
extinction of a lever-pulling response as functions
of percentage of reinforcement and amount of reward.
Journal of Experimental Psychology, 1957, 5£, 115-
120.
Lewis, D., and Duncan, C. P. Expectation and resistance to
extinction of a lever-pulling response as a function
of percentage of reinforcement and number of acqui
sition trials. Journal of Experimental Psychology,
1958, 55, 121-127.
Lindquist, E. F. Design and analysis of experiment in
psychology and education. Boston: Houghton Mifflin
Company, 1953.
Logan, F. A. Incentive. New Haven: Yale University Press,
1960.
Longstreth, L. The relationship between expectations and
frustration in children. Child Development. 1960,
31, 667-671.
Longstreth, L. Partial reinforcement effect and extinction
as a function of frustration and interference.
Journal of Experimental Psychology. 1964, 6 7 . , 481-
586.
Longstreth, L. Frustration effects in retardates with a
j secondary reinforcement paradigm. Paper read at
Society for Research in Child Development,
Minneapolis, March, 1965.
Madison, H. L. Experimental extinction as a function of
number of reinforcements. Psychological Reports,
1964, 14, 647-650.
!
Malamud, N. Neuropathology. In H. A. Stevens and R. Heber
(eds.), Mental retardation. Chicago: University of
Chicago Press, 1964. Pp. 429-452.
Malpass, L. F. Motor proficiency in institutionalized and
non-institutionalized retarded children. American
Journal of Mental Deficiency, 1960, 64, 1012-1015.
Malpass, L. F. Motor skills in mental deficiency. In N.
Ellis (ed.), Handbook of mental deficiency. New
York: McGraw-Hill Book Co., 1963. Pp. 602-631.
117
Marx, M. Some relations between frustration and drive. In
M. R. Jones (ed.), Nebraska symposium on motivation
1956. Lincoln: University of Nebraska Press, 1956.
Pp. 92-130.
McCoy, D. F., and Marx, M. H. Competing responses and the
partial reinforcement effect. Journal of Experi
mental Psychology. 1965, 7 0 . * 352-356.
McHose, J., and Ludvigson, H. W. Frustration effect as a
function of drive. Psychological Reports. 1964, 14.
371-374.
i
I
Mowrer, 0. H., and Jones, H. M. Habit strength as a func
tion of pattern of reinforcement. Journal of
Experimental Psychology, 1945, 3 5 . , 293-311.
Page, E. B. Ordered hypotheses for multiple treatments: A
significance test for linear ranks. Journal of the
American Statistical Association, 1963, 513, 216-230.
Peckham, R. H., Jr., and Amsel, A. Magnitude of reward and
the frustration effect in a within-subjects design.
Psychonomic Science. 1964, i, 285-286.
jPenny, R. K. The effect of non-reinforcement on response
strength as a function of number of previous rein
forcements. Canadian Journal of Psychology, 1960,
14, 206-215.
i
Perry, S. L., and Moore, J. W. The partial-reinforcement
effect sustained through blocks of continuous rein
forcement in classical eyelid conditioning. Journal
of Experimental Psychology, 1965, 69, 158-161.
i
Ross, R. Positive and negative partial reinforcement ex-
| tinction effects carried through continuous rein
forcement, changed motivation and changed response.
i Journal of Experimental Psychology. 1964, 6 8 . , 492-
502.
Ryan, T. J., and Cantor, G. N. Response speed in children
as a function of reinforcement schedule. Child
Development, 1962, 33, 871-878.
I
Ryan, T. J. The effects of nonreinforcement and incentive
value on response speed. Child Development, 1965,
36, 1067-1081.
Screven, C. G. The effects of interference on response
strength. Journal of Comparative & Physiological
Psychology, 1954, 47, 140-144.
118
[Screven, C. G., and Cummings, L. The effect of nonreward
I and interference on variation in the amplitude of an
instrumental response. Journal of Comparative &
Physiological Psychology, 1955, 48. 299-304.
Semler, I. Selective learning in severely retarded children
as a function of differential reaction to nonreward.
Child Development. 1965, 6 3 . , 143-151.
Sheffield, F. D., and Campbell, B. A. The role of experi
ence in the "spontaneous activity of hungry rats."
Journal of Comparative & Physiological Psychology.
1954, £7, 97-100.
Sheffield, V. F. Extinction as a function of partial rein
forcement and distribution of practice. Journal of
Experimental Psychology, 1949, 3 9 . ' 511-526.
Skinner, B. F. The behavior of organisms. New York:
Appleton-Century-Crofts, 1938.
^Spence, K. W. Behavior theory and conditioning. New Haven:
Yale University Press, 1956.
Spence, K. Behavior theory and learning. New Jersey:
Prentice Hall, 1960.
Steigman, M. J., and Stevenson, H. The effect of pretrain
ing reinforcement schedules on children's learning.
Child Development, 1960, 3 1 . , 53-58.
Super, D. E. Appraising vocational fitness. New York:
Harper and Brothers, 1949.
Theios, J. The partial reinforcement effect sustained
through blocks of continuous reinforcement. Journal
of Experimental Psychology, 1962, 64, 1-6.
juyeno, E. T. Effect of frustrative blocking on motivation.
Psychological Reports, 1965, 16. 203-208.
Wagner, A. R. The role of reinforcement and nonreinforce
ment in an "apparent frustration effect." Journal
of Experimental Psychology, 1959, 57, 130-136.
Weinstock, S. Resistance to extinction of a running re
sponse following partial reinforcement under widely
spaced trials. Journal of Comparative & Physiologi
cal Psychology. 1954, 47, 318-323.
119
jwilson, W., Weiss, E. J., and Amsel, A. Two tests of the
Sheffield hypothesis concerning resistance to
extinction, partial reinforcement, and distribution
of practice. Journal of Experimental Psychology.
1955, 50, 51-60.
Yates, A. J. Frustration and conflict. New York: John
Wiley and Sons, Inc., 1962.
APPENDIXES
120
APPENDIX A
PARENTAL CONSENT LETTER
121
MANHATTAN BEACH CITY SCHOOL DISTRICT
Dear _____________ ,
We are asking a few of the students within the
district to take a Manuals Skills Test. Basi
cally it involves boys' ability to do simple
manual tasks quickly and persistently.
If you have no objections, we will be asking
______________ to participate within the next
two or three weeks. This is something he should
enjoy.
Should you have any questions please contact
this office.
Sincerely,
Robert T. Pennington
Director of Special Services
J. Byron Burgess
Assistant Superintendent-Curriculum
RTP/ac
122
APPENDIX B
STATISTICAL TABLES
123_
124
TABLE A
MEAN TIME PER TRIAL IN SECONDS FOR PERFORMANCE
OF SECTION TWO OF TASK I— ADJUSTED MEANS
FOR RETARDATES IN PARENTHESES
Trial
Normals Retardates
Continuous
Reward
Partial
Reward
Continuous
Reward
Partial
Reward
1 7.98 8.20 8.63 (9.04) 10.50 (10.11)
2 7.43 7. 04 8.63 (9.04) 9.37 ( 8.98)
3 7.22 7.07 8.21 (8.62) 9.01 ( 8.62)
4 7.19 6.88 7.91 (8.32) 9.43 ( 9.04)
5 6.73 7.02 8.05 (8.46) 8.66 ( 8.27)
6 6.67 6.85 8.17 (8.58) 8.68 ( 8.29)
; 7 6.69 6.74 8.07 (8.48) 8.46 ( 8.07)
8 6.44 6.98 8.06 (8.47) 8.14 ( 7.75)
9 6.45 7.07 7.47 (7.98) 8.14 ( 7.75)
10 6.31 6.56 7.75 (8.16) 8.22 ( 7.83)
11 6.51 6.35 7.63 (8.04) 8.11 ( 7.72)
12 6.43 6.49 7.45 (7.86) 7.78 ( 7.39)
125
TABLE B
INTELLIGENCE GROUPS X TREATMENTS GROUPING
ANALYSIS OF VARIANCE OF TRIAL 3
OF SECTION TWO ON TASK I
Source of
Variation
Sum of
Squares
d.f.
Mean
Square
F-ratio P
(A) Intelligence
level
848,208 1 848,208 17.80 <•01
(B) Treatment
grouping
3, 643 1 3,643
A X B 14,272 1 14,272
Within treatments 2,478,610 52 47,665
Total 3,344,733 55
Per cent points for distribution of F
5% for 1 and 50 d.f. =4.03
1% for 1 and 50 d.f. =7.17
126
TABLE C
TREND ANALYSIS OF TRIALS 4 THROUGH 12
OF SECTION TWO ON TASK I
■ Source of
" Variation
Sum of
Squares
d.f.
Mean
Square
F-ratio P
Intelligence 9,990,999 1 9,990,999 30.47 <.01
Reinforcement 66,171 1 66,171
Intell. X
Reinforce.
136,982 1 136,982
Error (a) 17,049,527 52 327,875
fTrials 1,061,581 8 132,697 7.20 <.01
Intell. X Trials 122,443 8 15,305
Reinforce. X
Trials
175,318 8 21,914 1.18 >.05
Intell. X
Reinforce. X
Trials
276,423 8 34,552 1.87 >.05
Error (b) 7.665.719 416 18,427
Total 36,545,163 503
Per cent points for distribution of F
5% for 1 and 50 d.f. =4.03
1% for 1 and 50 d.f. = 7.17
5% for 8 and 400 d.f. =1.96
1% for 8 and 400 d.f. = 2.55
127
| TABLE D
j
TREND ANALYSIS OP TRIALS 4 THROUGH 12 OF SECTION TWO
ON TASK I FOR NORMAL BOYS
Source of
Variation
Sum of
Scruares
d.f.
Mean
Souare
F-ratio P
Reinforcement 6, 369 1 6, 369
Error (a) 11,916,095 26 458,311
Trials 610,497 8 76,312 3.61 <.01
Reinforce. X
Trials
226,482 8 28,310 1.34 >.05
Error (b) 4.391.751 208 21,114
Total 17,151,194 251
_ _
Per cent points for distribution of F
5% for 1 and 26 d.f. =4.22
I% for 1 and 26 d.f. =7.82
5% for 8 and 200 d.f. =1.98
1% for 8 and 200 d.f. =2.60
TABLE E
TREND ANALYSIS OF TRIALS 4 THROUGH 12 OF SECTION TWO
ON TASK I FOR RETARDED BOYS
Source of
Variation
Sum of
Squares
d.f.
Mean
Square
F-ratio P
Reinforcement 196,784 1 196,784
Error (a) 5,133,432 26 197,439
Trials 573,527 8 71,690 4.55 <.01
Reinforce. X
Trials
225,259 8 28,157 1.79 >.05
Error (b)
Total
3,273.968
9,402,970
208
251
15,740
Per cent points for distribution of F
5% for 1 and 26 d.f. =4.22
1% for 1 and 26 d.f. =7.82
5% for 8 and 200 d.f. =1.98
1% for 8 and 200 d.f. = 2.60
128
TABLE F
MEAN TIME PER TRIAL IN SECONDS FOR PERFORMANCE
OF SECTION ONE OF TASK I— ADJUSTED MEANS
FOR RETARDATES IN PARENTHESES
Trial
Normals Retardates
Continuous
Reward
Partial
Reward
Continuous
Reward
Partial
Reward
1 9.01 8.41 10.04 (10.42) 11.79 (11.39)
2 7.81 7.95 9.27 ( 9.65) 10.22 ( 9.82)
3 7.33 7.14 8.80 ( 9.18) 9.66 ( 9.26)
4 7.36 7.03 8.34 ( 8.72) 9.52 ( 9.12)
5 7.08 6.86 8.30 ( 8.68) 9.89 ( 9.49)
6 6.85 7.16 8.30 ( 8.68) 9 . * , 3 ( 9.09)
7 7.41 6.89 8.31 ( 8.69) 8.77 ( 8.37)
8 6.75 6.92 8.20 ( 8.58) 9.20 ( 8.80)
9 6.80 7.00 8.24 ( 8.62) 8.12 ( 7.72)
10 6.55 6.35 8.23 ( 8.61) 8.73 ( 8.33)
11 6.43 6.46 7.70 ( 8.08) 8.18 ( 7.78)
12 6.41 6.41 7.72 ( 8.10) 8.06 ( 7.66)
129
TABLE G
INTELLIGENCE GROUPS X TREATMENTS GROUPING
ANALYSIS OF VARIANCE OF TRIAL 3
OF SECTION ONE ON TASK I
Source of
Variation
Sum of
Squares
d.f.
Mean
Square
F-ratio P
(A) Intelligence
level
1, 340,445 1 1,340,445 41.80 <.01
(B) Treatment
grouping
895 1 895
A X B 10,477 1 10,477
Within treatments 1.667.488 52 32,067
Total 3,019,305 55
Per cent points for distribution of F
5% for 1 and 50 d.f. = 4.03
1% for 1 and 50 d.f. =7.17
130
TABLE H
TREND ANALYSIS OF TRIALS 4 THROUGH 12
OF SECTION ONE ON TASK I
Source of
Variation
Sum of
Squares
d.f.
Mean
Square
F-ratio P
Intelligence 12,477,477 1 12,477,477 54.00 <.01
Reinforcement 84,838 1 84,838
Intell. X
Reinforce.
24,431 1 24,431
Error (a) 12,014,102 52 231,040
i
Trials 1,676,527 8 209,565 12.25 <.01
Intell. X Trials 195,767 8 24,470 1.43 >.05
iReinforce. X
Trials
232,436 8 29,054 1.69 >.05
Intell. X
Reinforce. X
Trials
262,679 8 32,834 1.92 >.05
1
Error (b) 7.113.852 416 17,100
Total 34,082,109 503
Per cent points for distribution of F
5% for 1 and 50 d.f. =4.03
1% for 1 and 50 d.f. = 7.17
5% for 8 and 400 d.f. = 1.96
1% for 8 and 400 d.f. = 2.55
131
TABLE I
TREND ANALYSIS OF TRIALS 4 THROUGH 12 OF SECTION ONE
ON TASK I FOR NORMAL BOYS
Source of
Variation
Sum of
Squares
d.f.
Mean
Square
F-ratio P
Reinforcement 100,161 1 100,161
Error (a) 10,166,665 26 391,025
Trials 1,047,898 8 130,998 8.46 <.01
Reinforce. X
Trials
202,566 8 25,320 1.64 >.05
Error (b) 3.217.400 208 15,468
Total 14,734,781 251
Per cent points for distribution of F
5% for 1 and 26 d.f. =4.22
1% for 1 and 26 d.f. = 7.82
5% for 8 and 200 d.f. =1.98
1% for 8 and 200 d.f. =2.60
TABLE J
TREND ANALYSIS OF TRIALS 4 THROUGH 12 OF SECTION ONE
ON TASK I FOR RETARDED BOYS
Source of
Variation
Sum of
Squares
d.f.
Mean
Square
F-ratio P
Reinforcement 9,107 1 9,107
Error (a) 1,847,438 26 71,055
Trials 824,305 8 103,038 5.50
<•
01
Reinforce. X
Trials
292,550 8 36,568 1.95
>.
05
Error (b) 3.896,451 208 18,732
Total 6,869,851 251
Per cent points for distribution of F
5% for 1 and 26 d.f. =4.22
1% for 1 and 26 d.f. =7.82
5% for 8 and 200 d.f. = 1.98
1% for 8 and 200 d.f. =2.60
132
TABLE K
ANALYSIS OF VARIANCE IN A 2 X 2 X 2 FACTORIAL DESIGN
OF TRIALS TO EXTINCTION ON SEGUIN FORM BOARD
AFTER LOG TRANSFORMATION
Source of
Variation
Sum of
Squares
d.f.
Mean
Square
F-ratio P
Treatments .90771552 7 .12967365 1.85 >.05
Within
treatments
3.36974911 48 .07020311
Total 4.27746463 55
(A) Intelligence
grouping
.02495702 1 .02495702
(B) Reinforcement
history
.26656081 1 .26656081 3.80 >.05
(C) Terminal
Seguin
response
.08702674 1 .08702674
A X B .1146816 1 .1146816
A X C .00011031 1 .00011031
B X C .38161213 1 .38161213 5.44 <.05
A X B X C .03276691 1 .03276691
ferror: within
treatments
3.36974911 48 .07020311
Total 4.27746463 55
Per cent points for distribution of F
5% for 7 and 48 d.f. =2.21
1% for 7 and 48 d.f. = 3.04
5% for 1 and 48 d.f. = 4.04
1% for 1 and 48 d.f. = 7.19
133
TABLE L
MEAN TIME PER TRIAL IN SECONDS FOR FIRST EIGHT TRIALS
ON SEGUIN FORM BOARD— ADJUSTED MEANS FOR
RETARDATES IN PARENTHESES
Normals Retardates
Trial
Continuous
Reward
on Task I
Partial
Reward
on Task I
Continuous
Reward
on Task I
Partial
Reward
on Task I
1 18.18 18.33 20.22 (20.79) 22.94 (22.38)
2 15.07 14.20 16.95 (17.52) 19.84 (19.28)
! 3
13.74 13.48 16.03 (16.60) 17.97 (17.41)
4 13.96 13.13 15.79 (16.36) 18.15 (17.59)
5 13.77 12.85 15.05 (15.62) 17.53 (16.97)
6 13.76 13.37 15.63 (16.20) 16.43 (15.87)
i 7 13.06 12.57 14.77 (15.34) 16.58 (16.02)
8 13.34 12.37 15.49 (16.06) 15.76 (15.20)
134
TABLE M
CORRELATION COEFFICIENTS FOR THE RELATIONSHIP
OF CHRONOLOGICAL AGE WITH TASK PERFORMANCE
Variables Normals Retardates
All
Groups
Task I performance index
age
with
-.08 -.13 -.09
Seguin performance index
age
with
-.21 -.16 -.14
(Trials to extinction with age -.03 -.01 -.08
Seguin performance index
trials to extinction
with
.06 .04 -.08
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
An Application Of A Two-Stage 'Attention' Model To Concept Formation In The Mentally Retarded
PDF
The Effects Of Sex, Assigned Therapist Or Peer Role, Topic Intimacy, And Expectations Of Partner Compatibility On Dyadic Communication Patterns
PDF
Effects Of Success And Failure On Impulsivity And Distractibility Of Three Types Of Educationally Handicapped Children
PDF
Effects Of Video Tape Feedback Versus Discussion Session Feedback On Group Interaction, Self Awareness And Behavioral Change Among Group Psychotherapy Participants
PDF
The Effect Of Dissonance In Self-Esteem On Susceptibility To Social Influence
PDF
The Effects Of Two Types Of Experimenter Intervention And Schedules Of Reinforcement On Verbal Operant Conditioning Of Affective Self-References
PDF
The Effects Of A Self Shock Procedure On Hallucinatory Activity In Hospitalized Schizophrenics
PDF
Verbal Reports Of Emotional States And Onsets And Offsets Of Conditioned Stimuli
PDF
The Effects Of Anxiety And Threat On Self-Disclosure
PDF
The Effects Of Nonreward Dissonance And Secondary Reward On Extinction And Attractiveness
PDF
The Effect Of Discriminability On The Partial Reinforcement Effect In Human Gsr Conditioning
PDF
Non-Specific Treatment Factors And Deconditioning In Fear Reduction
PDF
Prediction Of Therapeutic And Intellectual Potential In Mentally Retardedchildren
PDF
Confirmation Of Expectancy And Changes In Teachers' Evaluations Of Student Behaviors
PDF
The Effects Of Justice, Balance, And Hostility On Mirth
PDF
The Enhancement Of Eeg - Alpha Production And Its Effects On Hypnotic Susceptibility
PDF
Prognostic Expectancy Effects In The Desensitization Of Anxiety Over Invasion Of Body Buffer Zones
PDF
The effect of noise on intellectual performance as related to personality and social factors in upper division high school students
PDF
An Effect Of Verbal Conditioning On Nonverbal Behavior
PDF
The Effects Of Prior Part-Experiences On Visual Form Perception In The Albino Rat
Asset Metadata
Creator
Barron, Robert Francis
(author)
Core Title
Transfer Of The Partial Reinforcement Extinction Effect Across Tasks In Normal And Retarded Boys
Degree
Doctor of Philosophy
Degree Program
Psychology
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
OAI-PMH Harvest,psychology, clinical
Language
English
Contributor
Digitized by ProQuest
(provenance)
Advisor
Jacobs, Alfred (
committee chair
), De Nike, L. Douglas (
committee member
), Grings, William W. (
committee member
), McIntrye, Robert B. (
committee member
)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c18-120233
Unique identifier
UC11360025
Identifier
6706488.pdf (filename),usctheses-c18-120233 (legacy record id)
Legacy Identifier
6706488.pdf
Dmrecord
120233
Document Type
Dissertation
Rights
Barron, Robert Francis
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Tags
psychology, clinical