University of Southern California Dissertations and Theses

A Cognitive Interpretation Of Acquired Reinforcement Effects

(USC Thesis Other)

A Cognitive Interpretation Of Acquired Reinforcement Effects

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content COGNITIVE INTERPRETATION OF ACQUIRED
REINFORCEMENT EFFECTS
by
Darena Ann Bailey
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)
June 1971
BAILEY, Darena Ann, 1944-
A COGNITIVE INTERPRETATION OF ACQUIRED
REINFORCEMENT EFFECTS.
University of Southern California, Ph.D.,
1971
Psychology, general
I University Microfilms, A XEROXCompany, Ann Arbor, Michigan
THIS DISSERTATION HAS BEEN MICROFLIMED EXACTLY AS RECEIVED
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES. CALI FORNIA 9 0 0 0 7
This dissertation, 'written by
under the direction of Dissertation Com
mittee, and approved by all its members, has
been presented to and accepted by The Gradu
ate School, in partial fulfillment of require
ments of the degree of
Bailey.
D O C T O R OF P H IL O S O P H Y
Dean
D a te ..3 -4 -J. . 1
DISSERTATION COMMITTEE
Chairman
ACKNOWLEDGMENTS
Acknowledgments and thanks are gratefully made to all
who have rendered assistance in the collection of the mate
rial for this dissertation.
Thanks are especially owed to Dr. L. E. Longstreth for
his guidance and help in the planning and execution of this
study and for his patient, constructive, suggestions con
cerning the final draft of this dissertation.
Sincere appreciation is also extended to Dr. S. Madigan
whose assistance in the latter stages of revision proved to
be invaluable. Finally, a debt of gratitude is owed to
Dr. M. B. Marks for his encouragement and help throughout
the preparation of this dissertation.
ii
TABLE OP CONTENTS
Page
LIST OF TABLES.......................................... iv
LIST OF ILLUSTRATIONS................................. V
Chapter
I. INTRODUCTION..................................... 1
II. REVIEW OF SECONDARY REINFORCEMENT LITERATURE . . 4
III. REVIEW OF COGNITIVE LITERATURE.................. 11
IV. DEVELOPMENT OF THE EXPERIMENTAL DESIGN......... 18
V. METHOD........................................... 25
V Design
Subjects
Materials and Apparatus
Procedure
VI. RESULTS......................................... 35
VII. DISCUSSION....................................... 50
APPENDUM.................................................. 57
REFERENCES................................... 61
LIST OF TABLES
Table Page
1. Mean Scores of Sa As a Proportion of Total
Number of Choices in Each Quarter........... 36
2. Summary of the Analysis of Variance of
Choices of S a ................................. 37
3. Values of the Studentized Range Statistics
for Mean Scores of S a ........................ 40
4. Number of Choices of Sa When Ss Were Asked
the First and Second Time.................... 42
5. Number of Subjects Preferring Sa with the
Reason of Wanting to Play the First
Game A g a i n ................................... 44
6 . Number of Subjects Preferring Sa with the
Reason of Wanting to Play the First Game
Again When Asked the Second T i m e ........... 46
7. Number of Subjects Starting Criterion Trials
as Related to Previous Choices of Sa . . . . 47
8 . Number of Subjects Starting Criterion Trials
as Related to Number of Performance
T r i a l s ............................ 48
LIST OF ILLUSTRATIONS'
Figure Page
1. The Significant Interaction Effects Between
(a) Age and Replay; (b) Replay and Blocks;
and (c) Age and Blocks..............................39
2. Choices of Sa, So and Sn Over Blocks of
Trials for the Four Experimental Conditions ......41
CHAPTER I
INTRODUCTION
Reinforcement as a way of controlling behavior becomes
immeasurably more complex as one moves from animal to human
behavior. To a hungry animal, food is an effective rein
forcing stimulus. However, in the case of human beings, it
is difficult to say what will serve as a reinforcing stimu
lus. We do know that reinforcement extends beyond food and
water such that a child may be reinforced by almost anything
— a parent's smile, a teacher's nod, or a word of social
approval such as "good." The behavior of adults is often
reinforced by such intangibles as respect, admiration, or
the favorable regard of one's spouse or friends. The main
concern of this paper will be to investigate the underlying
mechanisms that contribute to the effectiveness of these
acquired reinforcers.
The typical textbook explanation of acquired reinforce
ment usually points to experiments in which food is paired
with a neutral stimulus such as a buzzer over a number of
trials until the buzzer becomes reinforcing in and of itself
(Hilgard, Atkinson, & Atkinson, 1971; Kagan & Havemann,
1968; Kimble, 1961). The buzzer or any object that becomes
1____________________________________
2
rewarding through association with a primary reinforcer is
called a secondary reinforcer (Sr). Then it is usually
stated that many of the things that human beings consider
rewards seem to acquire their reinforcement value in this
way. For example, mother becomes paired with food, home
becomes paired with mother, and in later years the home can
become associated with the community and with certain stand
ards of good citizenship for which the community seems to
stand. Thus, through secondary reinforcement we are pro
vided with an explanation of why people in a community
become good citizens (Kagan & Havemann, 19 68).
An interpretation of acquired reinforcement which has
not received as much attention is one which is based primar
ily on the cognitive processes of the individual. According
to this point of view, acquired reinforcers are best under
stood by de-emphasizing stimulus-response associations and
by stressing instead the individual's perception of rela
tionships and organization of knowledge. This explanation
assumes that neutral stimuli such as social evaluations do
not acquire the power to automatically control behavior
after being paired with rewards. Rather, by consistently
pairing a social evaluation with a reward, the individual
learns to expect that the social evaluation is likely to be
followed by a reward in the future. If information is ac
quired which contradicts this expectation, the social evalu
ation would lose its value, regardless of the number of
3
times it had been associated with a reward. Thus, the
effectiveness of the acquired reinforcer depends on how the
person perceives it in relation to his ultimate goals
(Hilgard, Atkinson, & Atkinson, 1970; Hill, 1968; Long-
streth, 1970a, 1970c).
In comparing these two approaches, we can see that the
main difference between them is that the secondary rein
forcement approach, as generally presented, postulates the
key variable in learned rewards to be the contiguity of the
original neutral stimulus with the primary reinforcer. The
cognitive approach, on the other hand, hypothesizes that the
cognitive processes of the individual serve to mediate the
effects of the reinforcement. Note that this is not to say
that all learning theorists have restricted themselves to a
view of secondary reinforcement which only considers the
past pairings important, but rather that prevalent concep
tions of Sr have largely ignored cognitive factors in
accounting for the effects of acquired reinforcers.
The main purpose of this study is to explore the limi
tations of a general Sr approach as an explanation of
acquired reinforcement effects and to show how a cognitive
approach is better able to account for these effects. In
order to consider these alternative points of view more
fully, the next two sections of this paper will be devoted
to a fuller explanation of each position and a review of the
relevant literature.
CHAPTER II
A REVIEW OF SECONDARY REINFORCEMENT LITERATURE
Before the concept of secondary reinforcement can be
adequately explained, one must first understand primary
reinforcement. Thorndike was one of the first to allude to
the concept of reinforcement when he explained his law of
effect. Briefly, the Law of Effect states that responses
which are followed by events satisfying to the organism are
more likely to be repeated than are responses followed by
discomfort to the organism (Kimble, 1961). Hull (1943)
modified Thorndike's explanation to include the concept of
drive or need reduction. His explanation, which he called
the law of primary reinforcement, states:
Whenever a reaction (R) takes place in temporal contiguity
with an afferent receptor impulse (s) resulting from the
impact upon a receptor of a stimulus energy (S) , and this
conjunction is followed closely by the diminution in a
need (and the associated diminution in the drive, D, and
in the drive receptor discharge, sD), there will result an
increment, (s R), in the tendency for that stimulus on
subsequent occasions to evoke that reaction. (1943,
p. 71)
Thus, according to Hull, drive reduction constitutes the
sole basis of primary reinforcement.
Skinner (1953), on the other hand, has preferred not tc
theorize about the basis of primary reinforcement. Instead,
_______________________________ 4______________________________
he has commented that in order to establish whether an eventj
is reinforcing, one need only "observe the frequency of a
selected response, then make an event contingent upon it and
observe any change in frequency. If there is a change, we
classify the event as reinforcing to the organism under the
existing conditions" (Skinner, 1953, p. 73).
Although Hull and Skinner differ about the importance
of drive reduction as the basis of primary reinforcers, they
are in essential agreement that a secondary reinforcer
acquires its reinforcing properties through association with
a primary reinforcer or through association with any rein
forcer. As Hull stated it, "the power of reinforcement may
be transmitted to any stimulus situation by the consistent
and repeated association of such stimulus situation with the
primary reinforcement which is characteristic of need reduc
tion. Moreover, after the reinforcement power has been
transmitted from this to one hitherto neutral stimulus, it
may be transferred from this to another neutral stimulus,
and so on in a chain or series whose length is limited only
by the conditions which bring about the consistent and
repeated associations in question" (Hull, 1943, p. 97).
Using the term conditioned reinforcement in place of second
ary reinforcement, Skinner described the acquisition of
reinforcing properties. "If each time we turn on a light,
we give food to a hungry pigeon, the light eventually
becomes a conditioned reinforcer" (Skinner, 1953, p. 76).
It is not difficult to understand how the Sr principle
could be used to account for many different kinds of behav
ior. Skinner (19 53) mentions just a few of the things which
are said to be conditioned or secondary reinforcers:
attention, social approval, affection, and submissiveness of
others. It is important to recognize that the Sr principle
is in danger of being used as a catch-all phrase for all
previously unexplained behaviors in much the same way as the
terms "instinct" and "learned drives" were used. As Long-
streth (19 67) explained, a common practice has been to lable
almost any event which follows a response, but is not a
primary reinforcer as a secondary reinforcer.
Despite the rather widespread acceptance of secondary
reinforcement as an explanation of much of behavior,
attempted empirical demonstrations of secondary reinforce
ment at the animal level have met with only limited success
(Hertel, 1966) and those at the human level with little or
no success (Hertel, 1966; Lundgren, 1970; Longstreth, 1966b&
1967). Since we are concerned primarily with human behavior
in this paper, a brief explanation of the problems in the
human Sr research literature will be presented.
A study by Leiman, Myers, & Myers (1961) is typical of
the kind of procedure usually involved in those studies
finding Sr effects in children. Subjects sit in front of a
panel with three lights. The lights come on randomly, one
at a time, and the Ss are instructed to push the button
underneath the light that turns on. If they do this cor- !
rectly, they receive a candy. For the secondary reinforce
ment (Sr) group, there is a buzzer paired with candy during
training and only the buzzer presented during extinction.
One of the control groups (C^) has the same training proced
ure as the Sr group but receives neither buzzer nor candy
during extinction. The other control group (C2) receives
only candy during training and only the buzzer during
extinction. The fact that the Sr group made more correct
responses during extinction than either or C2 is often
cited as evidence that the buzzer became an Sr.
However, another interpretation is that the buzzer is
eliciting the next response rather than reinforcing the pre
vious response. Since the buzzer preceded the "next"
response in training and was paired with a primary rein
forcer, it is possible that the buzzer became conditioned to
the following response and served as a cue to elicit that
response, as occurs in classical conditioning. The same
optimal interval of approximately a half a second between
the CS and the UCS has also been found to be optimal in Sr
studies between the primary reinforcer and the nehtral
stimulus. In addition, the CS-response interval of only a
few seconds is comparable to the neutral stimulus-response
interval in Sr studies, allowing for the neutral stimulus to
become conditioned to the response. For example, in the
Skinner box when the rat has learned to make the correct
response, the response occurs only a few seconds after the
Ifood and buzzer are presented. Considering the conditions
used to establish the buzzer as an Sr, end the contiguity
between the buzzer and the next response, it is possible
that the buzzer serves as a cue or a CS for the next
response. There is no reason to assume that in addition to
its cue properties, the buzz also had reinforcing effects.
In a different study (Myers, Craig, & Myers, 1961) the
authors may have realized this possibility when they stated,
"The Sr, when administered in extinction, seemed to release
the additional behavior in the chain ..." (Myers, et al. ,
1961, p. 771).
Aside from the problem of elicitation, another way in
which human Sr studies have been confounded is by failing to
test for inherent reinforcing effects of the stimulus that
is to be paired with the primary reinforcement. Conse
quently, when the "neutral" stimulus is presented during
extinction, one does not know whether the results are due to
the pairing of the stimulus with a reward or to the effects
of the stimulus itself. Some studies which have been con
founded in this way include Black and Thomas, 19 66; Ellis
and Pryer, 1957; Sidowski, Kass, and Wilson, 1965; Myers,
1960.
To go over just one example briefly, Sidowski, et al.,
(19 65) tested for Sr effects by pairing a red light with the
presentation of a penny during conditioning. In the ex-
. g
tinction phase of the experiment, Ss in the Sr group who
were presented with the light alone during extinction showed
significantly greater resistance to extinction than the
other extinction groups. However, for none of the groups
was the light ever presented alone during conditioning.
Hence, it is possible that the light had inherent reinforc
ing effects of its own, which were not acquired through
pairing with the penny.
For an overall review of these sources of confounding
and their presence in the Sr literature, the reader is
referred to Longstreth (1966b), Hertel (1966), and Lundgren
(1970) . The reviews also include studies which have failed
to find Sr effects (Kass, Wilson, & Sidowski, 1964; Long
streth, 1962; Donaldson, 1961; Estes, 1960; Hall, 1964) and
some which have found frustration rather than Sr effects
(Longstreth, 1966a,b).
Some authors apparently are aware of the discrepancy
between fact and fancy in the Sr literature. As Bolles
(1967) stated, "... some theorists seem to have been so
sager to find laboratory evidence of secondary reinforcement
that they have been willing to accept purported claims that
were simply not substantiated by the data" (Bolles, 1967,
p. 368) . Bugelski (1956) questions whether the Sr is any
thing more than an elicitor of behavior:
. . . the click in the Skinner box is but one of a total
pattern of stimuli associated with one phase of what is,
contrary to usual opinion, a complicated serial response
rather than a simple or elementary behavioral unit. The
10
click is part of the stimulation that is associated with
approaching the food cup which automatically puts the
animal in position for pressing the bar. To select out
the click and glamorize it into a "secondary reinforcer"
is totally unnecessary, gratuitous, and theoretically
harmful. (Bugelski, 1956, p. 271)
Ylyers, in his 19 58 review, concluded that secondary rein
forcement "is inadequately defined and inadequately demon
strated" and that before any theory is based upon the
axistence of Srs more research is needed.
It is hoped that this brief discussion of secondary
reinforcement has provided the reader with an idea of some
of the problems facing this concept and with an understand
ing of this author's hesitation to unequivocally accept sec
ondary reinforcement as the explanation of acquired rein
forcement. First, there has not been a clear,- example of
secondary reinforcement in either animal or human studies,
leading one to doubt the validity of the phenomenon. Sec
ond, the Sr concept has been misused to the point where one
finds the label "secondary reinforcer" being used for many
events which affect the frequency of a preceding response
and which cannot be attributed to primary reinforcement. In
the next section data will be presented which points to the
Importance of mediating cognitive processes by the individ
ual.
CHAPTER III ‘
A REVIEW OF THE COGNITIVE LITERATURE
As previously mentioned, the alternative position un
der consideration for explaining the effectiveness of
learned reinforcement is based primarily on cognitive fac
tors . This position maintains that a stimulus acquires
reinforcing properties to the extent that it is perceived
as instrumental in obtaining a reward. In other words, if
an individual inferred that social approval such as "good"
from someone was likely to have significance to his goals,
then that social reinforcement would be more effective than
if he did not infer significance to his goals. An example
by Longstreth (1969) illustrates more clearly how this in
terpretation might apply in child-adult interactions:
A child comes in promptly from play when called, washes
his hands, and exhibits good manners while eating supper,
not because these responses are well learned due to past
social approval, nor because he is automatically antici-
; pating :the usual, rewards with which social approval has
been paired, but because he wants to stay up late in the
evening and watch a special program on TV. His strategy
is that if his parents are in a "good mood" after supper,
as evidenced by positive social evaluation, they will be
more likely to grant his request than if they are in a
11
12
"bad mood." His hypothesis, therefore, is that if he per
forms those responses which elicit social approval (e.g.,
a "good mood"), he will be able to achieve his goal (which
had not necessarily been previously paired with social
approval at all). Thus he is a "good boy," because he has
rationally concluded that such behavior is the best option
under the present set of circumstances. Tomorrow is
another day, and may call for a different strategy.
(Longstreth, 19 69, p. 12)
One line of evidence that points out that what a person
thinks about social evaluation contributes to the effective
ness of that evaluation comes from the work of Belgian psy
chologist Joseph Nuttin, as reported by Greenwald (1966).
Nuttin's main contribution was to demonstrate conditions
under which the law of effect did not apply. He found that
when he told Ss that they were not going to be tested on a
task to which they had been exposed (closed task), they did
not recall responses rewarded by "right significantly more
than responses followed by "wrong." He points out that in
open tasks, in which Ss are told they are to be tested
again, the superior retention of rewarded responses may not
indicate an "automatic effect of reward on stimulus-response
connections," but rather it may indicate that the effect of
reward is an artifact of Ss' intent to learn. In other
words, Ss' expectations about being tested again may be
responsible for superior retention of items followed by
"right" since he knows these will appear again whereas the
responses followed by "wrong" will be dropped. Although
fluttin's work contains some methodological difficulties, his
contribution is nevertheless important in pointing out that
13
further investigation is necessary before one can label the
effect of social rewards such as "right" and "wrong" as
automatic and independent of Ss' expectations and inten-
!
tions.
Other investigators have been concerned with the
information-giving aspects of neutral stimuli paired with
reward. Parker and Nunnally (1966) performed an experiment
with children in which nonsense syllables were associated
with rewards. They found that in the test phase, the sub
jects expected the nonsense syllable previously paired with
a penny to lead to a future reward of a quarter. The au
thors hypothesized that this expectancy task concerned "cog
nitive functions, relating to the extent to which Ss judge
that a conditioned stimulus will actually be related to the
receiving of rewards" (Parker and Nunnally, 1966, p. 331).
The results of another study, in which nonsense syllables
were not actually paired with reward, but rather with the
anticipation of a future reward (Nunnally, Duchnowski, &
Knott, 1967), also were interpreted in cognitive terms in
the sense that the neutral stimulus became a cue for a
forthcoming reward. In other words, the neutral stimulus
was interpreted as having informative value in relation to a
future reward, rather than serving a strictly reinforcing
function in and of itself. A strictly reinforcing function
is taken to mean increasing the probability of the occur
rence of a response in a learning situation (Kimble, 1961).
14
In addition to the increasing amount of evidence show
ing the importance of cognitive processes on acquired rein
forcer effectiveness, there have also been two studies which
have been concerned specifically with testing the Sr
explanation of social approval against the cognitive inter
pretation. The first one by Bailey (19 70) involved present
ing the social approval of "good" or the mark of minus
("-") to 8-year-old children under the conditions of Future
Reward (FR) and No Future Reward (NFR) . Ss in the FR condi
tion were led to infer a future reward depending on the
number of "goods" or minus sighs they earned; they were told
that if they earned enough, they could have a prize. The
task consisted of pointing to one of a pair of pictures for
a total of ten pairs. One of the members of each pair had
an animal in it, the other did not. Whenever Ss pointed to
the non-animal picture, E said "good" or marked a "-" on SB
sheet of paper, depending upon the group the Ss was in.
The prediction from the cognitive point of view was
that Ss in the FR condition would learn to point to the non
animal pictures first regardless of whether they received
"good" or . If only the past pairings of the neutral
stimuli with rewards are considered, as the traditional Sr
approach has done, then the Ss should have learned signifi
cantly better in the "good" condition rather than the "-"
condition since "good" has presumably been paired more with
primary rewards than
_
The results of this study supported the cognitive
interpretation. Fourteen of the twenty Ss in the FR condi
tion responded correctly to all ten pairs of pictures within
ten trials, while only three of the twenty Ss in the NFR
condition did so. The nature of the feedback had no effect.
It was concluded that a powerful determinant of the effec
tiveness of feedback, be it social approval or otherwise, is
what 2 3 "makes of it," and that past pairings with other
reinforcers may be of minor importance.
The second study by Lundgren (1970) involving children
7 years old, consisted of a training phase and a testing
phase. In the training phase, whenever a certain colored
light came on (counterbalanced so that it was green for some
and yellow for others) it was rewarded by social approval
and in addition on every third reward light presentation,
the S was allowed to choose a prize. In the testing phase,
the procedure differed according to whether the was in the
Future Reward condition or the No Future Reward condition.
In the former condition, the E told the £ 3 that she would be
busy in the next room but that the could continue playing
the game by pressing one of the two light buttons whenever a
a tone went on. Furthermore, the S was led to believe that
making the previously rewarded light come on enough would
earn him more prizes. In the latter condition, the S was
led to believe that there were no more prizes but that he
should go ahead and play the game of turning off the buzzer
_ . l6
t
by pressing one of the two light buttons while the E was out
of the room. The cognitive position would predict Ss in the
Future Reward condition to choose the light paired with
reward significantly more than the non-rewarded light in the
test phase whereas the No Future Reward should show no
preference for one light over the other. The Sr prediction
would be that Ss in both groups should choose the rewarded
light significantly more than the non-rewarded light.
Again, the cognitive prediction proved to be correct.
In the test phase, there was no learning among the No Future
Reward subjects: they responded at chance levels in the
selection of the rewarded light. On the other hand, the
subjects in the Future Reward group did learn to choose the
previously rewarded light. It appears then.that what is
important in determining the effectiveness of a learned
reward, is not strictly the pairings of that learned reward
with a primary reinforcer since this was the same for both
groups in the training phase but rather the significance of
that learned reward to the Ss' goals; in this case, whether
the learned reward was integrated into his strategy of
obtaining a prize.
Although it may be argued that there are alternative
interpretations to the studies presented which are in keep
ing with the Sr approach, we are dealing here with the way
the Sr concept has predominantly been used in the literature
with its emphasis on the pairings of a neutral stimulus with
17
a reward. Myers (1958) documents this fact in a review of
Sr experiments. The variables considered all had to do with
some aspect of the pairing situation: schedules of rein
forcement, frequency of primary reinforcement, temporal
intervals, and amount of primary reinforcement. Nowhere was
there any mention of -incentive, motivational, or cognitive
variables. In light of this fact, it was considered appro
priate to make predictions concerning the Sr approach on the
basis of the number of pairings that the neutral stimuli had
with the primary reinforcement. Results not explainable in
this manner were said to require an additional hypothesis
concerning the cognitions of the individual.
CHAPTER IV
DEVELOPMENT OF THE EXPERIMENTAL DESIGN
«
If a stimulus is reinforcing to the extent that it is
perceived as instrumental in the attainment of a reward,
the implication is that both the cognitive ability of the
E and the information available to the S. should be important
in determining expectancies of a reward from a stimulus.
Hence, cognitions and information will be discussed as they
relate to the development of the experimental design.
Piaget (1928) has provided a framework from which we
may be able to get some idea about the development of cog
nitions in the child. Piaget postulated a preoperational
period which includes ages 2 - 7 years and concrete opera
tional period from 7-11 years. According to Piaget, it
is not until the concrete operational period that the child
is capable of thinking logically. He reports that before
the age bf 7 there seems to be no pronounced desire for
logical justification, but that around the age of 7 the
child begins increasingly to use words such as "because"
and "since," to use “but" and "then" in a logical sense
and to have a genuine awareness of a need for a causal
agent.
19
Empirical data have supported Piaget's hypothesis of
the growth of logical thinking. In order to determine
whether the ability to infer increases with age, Kendler and
Kendler (19 62) tested 5- and 8-year-old children on an
experimental paradigm in which the S is trained on three
discrete behavior segments (A-B, X-Y, B-G) and. then in the
test situation is instructed to get G when only A and X are
available. They found that not : only did older subjects do
significantly better on this inference task, but also their
approach was more direct whereas the young Ss who solved the
problem correctly predominantly did so through indirect
means, i.e., " the goal achieved after the repetition of pre
viously acquired but presently irrelevant behavior segments"
(Kendler & Kendler, 1962, p. 465). This supports the idea
that inferential processes develop between the ages of 5 and
8.
In another study involving Ss from 3 to 20 years, Weir
(1968) also found evidence to support the idea that cogni
tive mediation increases with age. He found that the 3- to
5-year-old Ss responded to the task on a simple reinforce
ment notion only, rather than employing complex hypotheses
involving patterns of response as the older Ss did. For a
review of many other studies which have investigated the
intellectual changes that occur between the ages of 5 and 7,
the reader is referred to White (1965).
If the findings regarding the development of intel
20
lectual functioning are correct, then we should be able to
make certain predictions. We should be able to predict, for
example, that by varying the informational value of the
stimulus situation, we should be able to produce correspond
ingly different results in Ss old enough to integrate the
information and use it to make inferences, whereas for
very young Ss, the added information should not make a dif
ference .
Longstreth (197d)a) investigated this hypothesis in an
experiment involving 6 year olds and 9 1/2 year olds. The
experiment consisted of two phases. In Phase I, £s were
told they would be playing a game and they would have a
chance to win prizes. All Ss played a game which went as
follows: S took a paper clip (Sa) from a container and
inserted it into a hole in a marble machine, then pushed a
button which, in conjunction with the paper clip insertion,
turned on a light. Then the S turned off the light by push
ing a lever to the left. Occasionally the lever press
response resulted in a marble being automatically ejected.
When this occurred, the S took the marble to a "pretend
store" and put it in a container at the counter. The E then
took the container and returned it to the £ with a penny or
a nickel (S+) and a piece of wood (Sn) inside. The £ was
told he could keep whatever was in the container and that he
^ould play the game again by starting at the beginning of
the sequence.
Phase I was terminated by Ss running out of paper j
i
!clips. Then half of the Ss received one kind of information;
land the other half another. To half of the Ss it w a s ■ ■ —— -
i |
lexplained that they would be able to win prizes by playing
the first game again if they played a second game correctly
(Replay condition). The other half of Ss were told they
couldn't play the first game again but that they would have
a chance to win prizes by playing a second game correctly
(No Replay condition).
Phase II of the experiment consisted of the second game
that Ss were told they could play. In this game, £s chose
between either a blue cup, a green cup, or a yellow cup.
One contained a paper clip, one contained wood, and a third
cup had nothing (So) in it. After ten trials, each £ was
asked which he wanted— the paper clip or the wood and why.
Phase II was continued until a cup was chosen 5 times in a
row or until the £ completed 20 trials.
It was predicted according to the cognitive position
that the different instructions would have an effect on the
older £s1 behavior but not on the younger £s. More specif
ically, it was expected that 9 1/2 year olds given the
information that they could play Game 1 again would learn to
choose Sa in Phase II. The reason is that they are able to
infer and they were provided with sufficient information
from which to infer that Sa was needed to play Game 1 again
and furthermore, that playing Game 1 was the way to win
prizes. The 9 1/2 No Replay condition £s, on the other j
; j
hand, had the ability to infer but since they did not expect
to play Game 1 again, it was predicted that they would show
|no particular preference for Sa or Sn. The different infor
mation given to the 6 year olds was predicted to have no
effect, since they presumably did not possess the cognitive
capacity to make inferences such as "If Sa, then prizes."
The Sr prediction was that Ss in all conditions should learn
to choose the cup containing Sn, since it was the stimulus
paired most closely with S+.
The results supported the cognitive hypothesis. The
9 1/2 Replay condition was the only one in which there was a
definite preference for Sa, with 75% of the Ss reaching cri
terion. The Sr prediction, on the other hand, was not sup-
pored, only one S out of 48 chose Sn five times in a row.
It is clear from this experiment that a neutral stimu
lus does not automatically acquire reinforcing power after
consistent pairings with an established reinforcer.
Instead, it appears that for an object to be reinforcing,
the S must perceive that it has some relationship to his
future goals. Furthermore, in order for the £ to perceive
such a relationship, he must possess the cognitive skills
and he must possess sufficient information. As the experi
ment demonstrated, without one or the other, the object does
not become an acquired reinforcer.
Although the basic logic of the preceding study is
................... ........ ................... 23
I i
sound, certain methodological problems exist: |
i ;
j (1) Probably the most serious problem is the fact that j
Sn was consistently paired with a reward while Sa was incon-j
I . '
sistently associated with rewards. That is, inserting Sa ini
the machine led to a marble only some of the time, whereas
each time Sn was presented, a marble was also presented.
Consequently, expectations to receive money from Sa were
partially reinforced, but expectations to receive money from
Sn were always reinforced. Therefore, it may have been that
Ss chose Sa more in the second game because the expectancies
to receive money from choosing Sa were more resistant to
extinction than were expectancies to receive money from the
Sn.
(2) There is also the problem that Ss1 preferences were
biased because Ss received and handled more Sas than Sns.
This allows for the possibility that familiarity with Sa may
have influenced Ss' choices to produce the obtained results.
For example, investigations have shown familiarity to be
reinforcing in some situations (Cofer & Appley, 1964). On
the other hand, it is also possible that Sjs were curious
about Sn which would have contributed to the opposite find
ings. Although there is no clearcut evidence as to the
relative strengths of familiarity or curiosity in this
situation, it would be desirable to control for the possi
bility that either or both might have been operating.
(3) Another uncontrolled variable was the sequence in
'................... 24
which Sa and Sn were presented. Sa was always received
ifirst in the sequence and Sn second. Although this invest!-
| |
jgator is not aware of any data indicating a strong likeli
hood that sequence effects were operating, this appears to
be another area where controls would be advantageous and
relatively simple to institute.
(4) Another problem was that an empty cup (So) was used
to evaluate whether Sn acquired any secondary reinforcer
effects with the idea that the extent to which Ss chose Sn
over So was a measure of Sns value as a secondary reinforcer,
However, it is possible that Ss would have chosen a cup con
taining any object over a cup containing nothing, regardless
of whether the cup containing the object had been paired
with a reinforcer. Consequently, there is no proper control
in the study to evaluate Sn as a secondary reinforcer.
(5) A further problem concerns the fact that Ss were
asked which object they wanted and why they wanted it after
10 test trials had passed. The problem is that Ss may have
been responding to the question on the basis of their
choices in the first 10 trials rather than according to the
one they "really" liked the best. Since the cognitive posi
tion is that Ss formulate hypotheses and then develop
preferences which reflect their hypotheses, getting an
accurate estimate of the Ss1 preferences would appear to be
a necessity for a cognitive interpretation of the data.
Because Longstreth's study is an important one in test-
ling the cognitive position, the present study is a pseudo- j
! I
replication of that study, with corrections for problems ;
I i
I
previously discussed. First, the problems with Sa being
inconsistently associated with rewards, more familiar to the
S, and first in the sequence were handled in the following
manner.
It was decided to deliver a marble to the S each time
he inserted Sa into the machine during the learning trials.
The only exception to this rule was on the second instruc
tional trial at which time delivery of a marble was not
given to avoid interference with the S learning the se
quence. To correct for the problem of familiarity, both Sa
and Sn were received by the Ss an equal number of times
making Sa and Sn equally familiar. Removal of sequence
effects was handled by the S receiving Sa at the same time
as he received Sn, again with the exception of the second
and third trial at which time it was necessary to have the JS
receive Sa first to begin the sequence.
The problem of evaluating Sn as a secondary reinforcer
was handled by having Ss choose between Sa, Sn, and a third
cup containing a neutral object. In order to control for
the effects of familiarity, the neutral object was presented
to the S at random intervals to Phase I with the same fre
quency as the Sa and the Sn. Therefore, the object Sn and
the neutral object So were equal in all respects except that
Sn was paired with a reward and So was not. Consequently, a
preference for Sn rather than So would constitute support
for the idea that Sn had become a secondary reinforcer.
i |
Finally, in order to obtain Ss1 unbiased preferences i
for the objects, Ss were asked before they began their test
trials which object they preferred. The idea here was to
find out if Ss1 verbal preferences before beginning the test
phase would correspond to their actual choices. If it did,
we could be certain that preferences guided their behavior
and not vice versa. Ss were asked for their choices two
other times. One time was at the very beginning of the
axperiment, before the first game was even explained. The
other time was after £s had completed 10 test trials, in
order to determine whether Ss preference for the objects
changed after they had an opportunity to choose among the
objects.
Thus, the purpose of the present experiment was to
investigate the conditions under which a stimulus acquires
reinforcing value. Is a stimulus likely to acquire rein
forcing value when it not only has been associated with an
established reinforcer, but is also instrumental in obtain
ing the reinforcer (e.g., Sa), when it is paired spatially
and temporally with an established reinforcer, but is not
instrumental in its obtainment (e.g., Sn), or when it is
neither paired with nor instrumental in the obtainment of a
reinforcer, but is familiar to the S (e.g., So)? The Sr
prediction is that the condition in which a stimulus is
n......... ' ........... ~... ' ... ‘ 27
closely paired with an established reinforcer, both spatial
ly and temporally, is the one in which the object should
jacquire reinforcing value. The cognitive prediction, on the
|
I
other hand, is that it is the condition in which a stimulus
is instrumental in obtaining a reward that the stimulus
should become reinforcing, but only to Ss who are able to
perceive the instrumental function of the stimulus.
The details of the experiment are described in the next
section.
CHAPTER V
METHOD
Design
A 2 x 2 x 4 factorial design was employed with the
variables being age (six and ten years), instructions (£s
were told that they could or could not return to the first
game), and blocks of trials (trials were divided into
quarters for each S).
The main dependent variable measure was the frequency
with which Ss chose cups containing objects Sa, Sn, or So.
Sa was the object both paired with the reward and instru-
mental in getting the reward, Sn was the object paired with
the reward, and So was the object neither paired with the
reward, nor instrumental in obtaining a reward.
Subjects
Subjects were obtained from an elementary school in the
Los Angeles area. An equal number of males and females froir
each of two age groups served as Ss with twenty-four 10-
year-olds and twenty-four 6-year-olds assigned to the two
treatment conditions with the restriction that each treat
ment group contain an equal number of males and females.
2 8 ________________________________
' "....... ' 29
The mean ages were: Male— 6 yr., 4 mo. and 10 yr., 2 mo.;
Female— 6 yr., 2 mo. and 10 yr., 3 mo.
Materials and Apparatus
The experiment was conducted in a mobile laboratory
located on the school grounds. The apparatus used was a
marble machine which consisted mainly of a stimulus-response
unit and a control unit. The S-R unit was situated in one
room of the trailer and the control unit was in the adjacent
room of the trailer with a one-way mirror between the two
rooms. Below the one-way mirror, there was a small open
window which was used by £s in this experiment to trade in
objects for money.
The stimulus-response unit of the marble machine con
sisted of a large black box, 44 in. high and 34 in. wide.
It rested on a low table so that a 9 in. square milk glass
window mounted on the front surface was even with £s eye
level. Directly below the window, a lever protruded which
could be pushed down to the left. To the right of the win
dow was clear plastic tube which received marbles automatic
ally ejected into it on a programmed basis. In the present
experiment, the marble reinforcements were programmed so
j
that a marble was ejected whenever the lever was pressed,
with the exception of two instructional trials which yielded
no marbles.
I Procedure
|
Phase I !
" I
The experimenter worked with each S individually. Upon
S's arrival at the trailer, E explained that they were going
to play some games and that S would have a chance to win
some prizes. The experimenter pointed to several gift-
wrapped prizes which were on the floor near the marble
machine. Then E showed three objects to the S (a piece of
wood, a paper clip, and a tack) and asked which he pre
ferred. S's preference was recorded on the data sheet.
The E then went to the control unit, and pulled the
toggle switch up to "on," turning on the bright light in the
S-R unit. E called S's attention to the light, saying, "See
the light? You can turn it off by pushing this lever." E
pushed the lever and turned off the light, saying to the S,
"See, it's off now." You can turn the light back on again
by taking this paper clip and putting it in the machine."
E then showed S how to put the paper clip in the ma
chine and how to press a button on the control unit at which
time the light on the S-R unit went on. Then E remarked,
"Now you have the light on and I will let you turn it off.
Before you do that, sometimes when you press the lever to
turn the light off, you will get a marble. There is some
thing very special you can do with the marble when you get
one. Go ahead and play the game and see if you can get a
marble." S then pressed the lever but no marble was deliv-
;........... " ..... '......... ' ” 31
i
!ered on this second instructional trial.
i S continued the sequence by activating the light in the
manner described and pressing the lever again. All subse-
I
j
quent lever presses resulted in a marble. £ 3 was told to
|
bring the marble over to the open window which E referred to
as the "pretend store" window. E said "Put the marble in
the cup and that means you want to buy something. I'll go
around inside and sell you something. Whatever I sell you,
you get to keep."
E went into the other room, removed the marble from the
cup and gave back to the S two cups. One contained a tack
and money (penny or nickel) . Both pennies and nickels were
used for variety in order to sustain Ss' interest in the
game. The other cup contained a paper clip. E suggested to
S that he take the tack and money and save them on a stand
in the corner of the room and that he leave.the paper clip
over by the machine where it could be used to play the game
jagain. E told S that if he collected enough money, he would
be able to buy the prize. S was asked if he understood the
I
jgame and was asked to explain it to E. His answer was
recorded on the data sheet. Ss then continued playing the
same while E monitored. At random intervals, E gave the £ a
piece of wood. S became familiar with it by receiving it
the same number of times as the tack and the paper clip, but
anlike those objects, it was neither paired with the reward
lor did it lead to a reward.
.. ......~..' .....................' .' ... 32
! Ss earned a total of six marbles, each traded for a
i
penny or a nickel and tack in the following sequence: P -
N-P-N-N-P. This sequence rather than a strictly
alternating one was used to arouse Ss1 curiosity since he
could never be sure whether to expect a penny or a nickel.
After the last marble was received, the trade-in at the win
dow was modified. This time there was no paper clip in the
cup. When saw that it was missing, E said Comment A or
Comment B, described below.
Comment A: No Replay (NR) treatment condition.
"You're out of paper clips. Well, that's all for that game.
We aren't going to play that game anymore. I'll turn off
the machine now." E turned off the machine and joined Ss.
E told S that there wasn't enough money to buy a prize
and then said, "Now let's play another game. If you play it
right, you'll be able to win a prize."
Comment B: (Replay [R] treatment condition). "You're
out of paper clips. We can't play this game anymore. If
you had gotten enough money, you could have bought one of
these prizes here, but you don't have enough money.
'Now let's play another game. If you play it right,
you'll be able to play this game again."
Phase II
All £s were then given an opportunity to choose between
the paper clip, the tack, and the wood in the following
manner. S was seated on a chair in front of the window. E
| 33
i
|
Was on the other side of the window and displayed 3 cups to
the S saying, "See these 3 cups? Well, I'm going to put a
i
apiece of wood in one of them. (E held up wood), a paper
clip in one of them (E held up paper clip), and a tack in
one of them (E held up tack). But you won't see me when I
do it, because I'm going to do it over here where you can't
see." E placed objects in cups out of S's vision. "Now you
name a color of a cup and you get to keep whatever is in it
and then give me the cup back. We'll play this game a few
times. I'll always put the things in the same color cups— I
won't switch them around." S^ was asked to explain this game
to the E and then asked what he wanted to win the most— the
wood, the paper clip, or the tack. His answers and the
reasons for his preferences were recorded. Ten trials were
administered with S's choice on each trial recorded.
After ten trials, S was asked again which he preferred
— the wood, paper clip, or tack and why. Then E reminded
the S that he would always put the things in the same cups
and that if he played the game right, he (Comment A) could
win one of the prizes or (Comment B) could play the other
game again. This game was continued until any one of the
objects was chosen 5 times in a row or until 20 trials had
been administered.
The objects were counterbalanced against conditions Sa,
Sn, and So to control for possible effects of objects, per
se. Each of the three objects— the paper clip, the wood,
34
and the tack were used an equal number of times for the Sa,
Sn, and So. To control for possible color effects, the
colors of the cups were assigned randomly for each S^ to the
three objects.
35
I
CHAPTER VI
RESULTS
The mean scores for choices of Sa are presented in
Table 1. Since Ss received an unequal number of trials,
scores were Vincentized so that scores are presented for
; the first, second, third, and fourth quarter of each S's
i
| total trials. In addition, mean scores of Sa are presented
l
as proportions of the total number of choices in a quarter
for each S_ since number of trials in quarters could vary.
A 2 x 2 x 4 factorial analysis of variance was applied to
the data with one factor being Age (six and ten years), In
structions (Replay and No Replay), and Blocks (trials were
divided into quarters for each S) . The results of this
analysis are presented in Table 2. The factor of sex was
not included in the analysis since the difference in mean
scores between the males and females was very small (X =.38
F
M
As shown in Table 2, all main effects and interac
tion effects were significant. The significant Age effect
indicates that there was a difference in the proportion of
__ch.oices of Sa according to the age of the S. Inspection
36
TABLE 1
Mean Scores of Sa As a Proportion of Total Number
of Choices in Each Quarter
Groups Quarters
1st 2nd 3rd 4th Row averages
Replay
Six years
.34 .33 .31 .43 .35
No
replay .30 .30 .34 .32 .32
Replay
Ten years
.31 .36 .61 .92 .55
No
Replay .36 .34
.28 .40 .34
Column averages .33 .33
.38 .52
of means in Table 1 shows the ten year old Ss had a greater
proportion of choices of Sa than the six year olds (X^q=.44
Xg=.33). The significant Replay effect indicates that the
proportion of choices of Sa differed according to whether
the S_ was in the Replay or No Replay condition. Inspection
of means shows the Ss in the Replay condition had a greater
proportion of choices of Sa than the Sjs in the No Replay
condition (X^=.45; XNR=.33). The significant Blocks ef-
TABLE 2
Summary of the Analysis of Variance of Choices of Sa
Source of Variation
Sum of
Squares df
Mean
Squares F
Between subjects
Age (A) .59 1 .59 28.25*
Replay (R) .72 1 .72 34.26*
A x R .33 1 .33 15 r66*
Subjects within groups .92 44 .02
i
I
Within subjects
Blocks (B) 1.11 3 .37 9.75*
A x B .56 3 .19 4.99*
R x B .77 3 .26 6.83*
A x R x B .54 3 .18 4.77*
B x subjects within
groups 4.99 132 .04
*p<.01
feet indicates that there was a difference in the propor-
tion of choices of Sa over blocks of trials, and as the
column averages in Table 1 show, the proportion of choices
of Sa increased over trials
The significant Age x Replay effect indicates that
I the proportion of choices of Sa was greatest when the ten
i
year olds were in the Replay condition as shown in Figure
1 (a). The finding of a significant Replay x Blocks effect
indicates the proportion of choices of Sa were greatest in
the Replay condition when £>s were in the later trials as
| shown in Figure 1 (b). The last significant two-way inter-
i
! action effect of Age x Blocks indicates that the proportion
i
of choices of Sa were greatest among the ten year old £>s in
I
|the later trials as shown in Figure 1 (c). The significant
| three-way interaction effect of Replay x Age x Blocks indi-
cates that choices of Sa were greatest in the Replay condi
tion when j3s were ten years old and in the later trials as
shown in Table 1 and Figure 2 (a)..
In order to determine whether the 10-year, Replay Ss
differed significantly from all the other groups and wheth
er any of the other groups differed from each other, a New-
man-Keuls test was performed (Winer, 1962, p. 80). In
Table 3 the values of the studentized range statistic are
presented. Inspection of Table 3 indicates that the only
group differing from the other groups was the 10-year,
Replay group.
M ean choices o f S a as a proportion o f total number of choices
1.00 1.00
.9 0 .9 0 .9 0
.8 0 . 8 0
.7 0 .7 0 .7 0
.6 0 . 6 0 .6 0
R
.5 0 .5 0 • 5 0
.4 0 .4 0 .4 0
.NR
NR
NR 10
NR —
.3 0 .3 0 .3 0
.20 .20 .20
.10 .10 .10
1st 2nd 3rd 4th 1st 2nd 3rd 4th 6 10
Age Quarters Quarters
FIG. 1 The Significant Interaction Effects between (a) Age and Replay; (b) Replay and Blocks; and
(c) Age and Blocks.
R = Replay, NR = No Replay
40
TABLE 3
Values of the Studentized Range Statistic
for Mean Scores of Sa
6, 6, 10, 10,
No replay Replay No Replay Replay!
______________________________________________I
j
6, no replay 1.44 .96 11.08*!
6, replay .48 9.63*;
!
10, no replay 10.20*1
10, replay
*p<.01
Vincentized learning curves are presented in Figure 2.
Trials are grouped in quarters and Ss1 choices of Sa, So,
and Sn in each condition are plotted. Again, we see by the I
marked upward trend of Sa in the 10-year, Replay that
i
choices of Sa in this group are significantly greater than |
in the other groups. Furthermore, in the 10-year, Replay
condition, but not in any of the other conditions, there is
a marked downward trend for Sn and So. Thus, the prediction
jthat only Ss in the 10-year, Replay group would learn to
choose Sa was supported.
In addition to £s' preferences of Sa as indicated by
jtheir performance in choosing Sa, Ss1 verbal preferences for
!
Sa during the experiment were also investigated. In Table 4
j
^re presented the number of Ss in the six-year-old group and
the ten-year-old group who chose Sa when asked the first
(a) (b) (c)
A 6 yr, No Replay
o 6 yr, Replay
■A 10 yr, N o Replay
« 10 yr, Replay
1.00
.9 0
.8 0
.7 0
60
.5 0
.4 0
.3 0
.20
.10
1st 2nd 3rd 4th
1.00
M
O
» .9 0
° .8 0
70
60
.5 0
« .4 0
M
0)
» .20
.10
1st 2nd 3rd 4th
1.00
.9 0
80
70
.6 0
50
4 0
.3 0
( / )
20
.10
1st 2nd 3rd 4th
BLOCKS OF TRIALS IN QUARTERS
FIG. 2 Choices of Sa, So, and Sn over blocks of trials for the four experimental conditions, where Sa = the
object instrumental in obtaining a reward; Sn =the object paired with a reward; and, So =the object neither
paired with nor instrumental in obtaining a reward.
TABLE 4
42
Number of Choices of Sa When Ss
Were Asked the First and Second Time
First time
Second time
No Yes
Six-year-old subjects
Yes 5 5
No 6 8
Ten-year-old subjects
Yes 2 6
No
3 13
time and second time. It was hypothesized that there would
be more _Ss answering Sa the second time than the first time
among the ten year old _Ss but not among the six year old
j3s. A chi-square test described in McNemmar (1962, p. 25)
designed to measure changes in frequency from the first ad
ministration to- the second administration was applied to
the data. Assigning the letter A to the numbers in the up
per left hand corners of the tables and the letter D to the
numbers in the lower left hand corners, the chi-square for-
2 9
: mula is X = (A-D) /A+D. Applying this chi-square to the
i
data in Table 4 yielded a significant chi-square for the
n
ten year olds (X =8.07, p<.025), but not for the six year
2
olds (X =.69, n. s.), supporting the hypothesis. Because
of the small number of observations, a chi-square test
could not be made between the ten year, Replay and ten
year, No Replay conditions.
Along with being asked their preferences of the ob-
| jects, Ss were also asked why they preferred the objects.
i
l
i
I It was predicted that not only would Ss in the ten year,
i
Replay condition learn to choose Sa, but also that they
would express a desire to win Sa before being given the
choices and that their reason for wanting Sa would be in
order to play the first game again. Table 5 shows the dis
tribution of Ss who, when asked their preferences the se
cond time, just before the test trials began, responded
with the answer Sa and the expected reason. Ninety-two
per cent (11 out of 12) of the ten year old Ss in the Re
play condition said they wanted to win Sa and gave as their
reason the desire to play Game 1 again, whereas in the No
Replay condition, only twenty-five per cent (three out of
12) responded in this manner. Pearson's chi-square (Hays,
1963, p. 589) as applied to these data was significant
44
TABLE 5
Number of Subjects
the Reason of
the First
Preferring Sa with
Wanting to Play
Game Again
Responding with Sa
Instructions
and the reason
Replay No replay
\
Six-year-
i
•old subjects
Yes 1 0
No 11 12
Ten-year-•old subjects
Yes 11 3
No 1 9
45
2
! (X =8.40, p<.005), indicating the number of Ss reaching
| criterion in the Replay condition was significantly greater
1
j
'than the number in the No Replay condition. For the six
year old £[s, no chi-square was performed because the ex
pected frequencies were extremely small; however, it is
clear from Table 5 that approximately an equal number of
six year old £^s in the Replay and No Replay conditions
I
i
I responded with Sa and the reason. The cognitive hypothesis
jwas again confirmed. Not only did ten year, Replay .Ss pre-
j
I fer Sa, as indicated by performance, but they were also
|
j
! able to give verbal evidence to the effect that they had
i
i
inferred why they should choose Sa, namely because that
was the object needed to play the first game again and
hence, to win prizes.
The next question that may be asked is whether there
is a relationship between reaching a performance criterion
of choosing Sa five times in a row and giving Sa slong with
the logical reason as the response. In other words, do the
same .Ss meet both the verbal and the performance criteria?
The data in Table 6 indicate that the answer to this ques
tion is affirmative. Eleven out of twelve of the j3s who
reached the performance criterion gave as their answer Sa
and the reason, whereas only one out of twelve of the Ss
TABLE 6
Number of Subjects Preferring Sa with the
Reason of Wanting to Play the First Game Again
When Asked the Second Time (Prior to Phase II)
Subjects Subjects giving Sa and
reaching Expected Reason
criterion
Yes No
Yes 11 1
No 4 32
I
not reaching criterion gave Sa and the reason. Pearson's
chi-square applied to the data was significant (X =23.56,
p<.001).
Another question concerns the learning pattern of the
J5s who reached the performance criterion. It is interest
ing to note that there did not appear to be any particular
pattern for reaching criterion. It might be expected that
Ss would choose Sa five times in a row upon discovering
the cup that contained Sa. However, only two Ss performed
in such a manner. Perhaps the reason why the other Ss did
not continue to choose Sa after discovering the cup con
taining Sa was that they wanted to find out what was in the
other two cups. If this were the case, then we would ex-
47
TABLE 7
Number of Subjects Starting Criterion Trials
as Related to Previous Choices of Sa
Experimental groups
Previous choices _______________________________________________
of Sa
6 year, 6 year, 10 year, 10 year,
replay ho replay replay no replay Total
0 2 2
1 3 3
2 1 1
3 1 1 2
4 2 1 3
5 0
6 1 1
Total number of subjects reaching criterion 12
pect the remaining ten of the criterion Ss to choose Sa
five times in a row after they had one previous choice of
Sa and a chance to try out all of the cups. However, only
three criterion Ss fell into this category. The rest of
the criterion Ss (seven) started their criterion trials
TABLE 8
Number of Subjects Starting Criterion Trials
as Related to Number of Performance Trials
Experimental groups
Performance trials---------------------------------------- total
6 year, 6 year, 10 year, 10 year,
replay no replay replay no replay
1-2 0
3-4 3 3
5-6 1 1
7-8 0
9-10 1 1
11-12 4 4
13-14 1 1 2
15-16 1 1
Total number of subjects reaching criterion 12
after having anywhere from two to six previous choices of
Sa with no apparent systematic differences as can be seen
in Table 7. The possibility was also explored that perhaps
more _Ss reached criterion as trials progressed. If this
were the case, we would expect to find a clustering of Ss
reaching criterion in the later trials. However, inspec
tion of the data in Table 8 reveals an apparent even dis
tribution of Ss reaching criterion over the performance
trials.
CHAPTER VII
DISCUSSION
The findings of the present study clearly support the
main hypothesis that objects acquire reinforcing value if
jthey can be expected to lead to future rewards. The predic-
!
jtion was that only those Ss who had both sufficient informa-
!
jtion from which to make the inference "If Sa, then prizes,"
|and also the cognitive ability to infer, would learn to
choose Sa as opposed to So or Sn. The 10-year-old Ss in the
Replay condition fulfilled both of these conditions and as
the statistical analyses revealed, they were the only ones
to show significant learning for the object Sa.
Subjects in the other three conditions did not show
Learning for any of the objects as predicted by the cogni
tive hypothesis. The 10-year-olds in the No Replay condi
tion had no logical reason to choose Sa since they did not
expect to be able to play Game 1 again and their receiving
prizes was dependent on E1s judgment of whether they played
Game 2 "correctly." On the other hand, the 6-year-olds in
the Replay condition did expect to play Game 1 again and
have a chance to win more money but they could not infer
that they should choose Sas to put in the machine in order
50 ________________________
to win more money. Subjects in the 6-year-old, No Replay
condition had neither cognitive ability nor sufficient
jinformation to make inferences and they were, of course, not
expected to learn to choose Sa.
These findings are clearly not in line with the Sr
hypothesis. According to the Sr point of view, objects
acquire reinforcing value through pairings with primary or
already established reinforcers. In the present study, the
object used as an established reinforcer was money. Previ
ous studies have shown that money in as small a value as a
penny to be reinforcing to both six-year-olds and ten-year-
olds (Longstreth and Hertel, 1970; Longstreth, 1970d). The
neutral stimuli paired with money were the Sa and the Sn.
iBoth were presented at the same time as the money and both
were close spatially although the Sn was slightly closer
spatially since it and the money were in the same cup while
the Sa was in an adjacent cup. Consequently, both the Sn
and the Sa should have acquired reinforcing properties and
should either have been chosen equally by the Ss, or, if
there was a preference, Sn should have been chosen more
because of its greater spatial contiguity to the reward. As
can be easily seen from the statistical analyses as well'as
from the learning curves, there was no evidence at all for
any preference of Sn. In fact, there was even a slight
preference among Ss for the So, the object never paired with
a reward, rather than the Sn. Two of the Ss choose So five
times in a row while none chose Sn five times in a row.
| A second prediction of the cognitive hypothesis was
that the 10-year-old, Replay Ss would be most likely to
express a preference for the object Sa when asked what they
wanted the second time and also to give the reason: to play
the first game again. This hypothesis was also confirmed by
the data. Hence, 10-year-old, Replay Ss were able to give
verbalizations consistent with their non-verbal behavior;
namely, that they had to choose Sa because that was the
object needed to play the first game again and to win
prizes.
Further analysis revealed that with four exceptions, it
was the same Ss who reached criterion that also gave as
their reason the desire to win Sa in order to play the first
game again. When questioning those four ^s who expressed a
desire to win Sa but who did not reach criterion with Sa,
they explained their behavior with answers such as, "I
thought it would be better to pick a little of each," or "I
didn't want to choose (Sa) all the time because it
seemed too hoggish." The other Ss who stated a preference
for Sa were consistent in choosing Sa on the performance
trials. Furthermore, they expressed their wish to win Sa
before choosing Sa five times in a row, supporting the cog
nitive position that behavior follows and is guided by
reasons, not vice versa.
It could be argued that perhaps the instructions were
less well understood by the first graders than by the fifth
graders and that this may account for differences in learn-
|
ing. However, a check was made on the experimental manipu
lation of instructions by asking each S if he understood the
game and then he was asked to explain it back to the E.
Thus, if and only if the S understood what had been
explained to him previously could he continue to the next
part of the experiment. As was expected, six-year-old Ss
jseemed to have little or no difficulty in understanding what
I
the E had said. This is not too surprising since the
instructions were relatively simple and probably easier to
understand than those they received in the regular school
activities.
Motivational factors were also considered for possible
effects on Ss1 behavior. It could be argued that there were
differences in motivation of the two age groups for the
money. If the differences were such that six-year-olds were
not as interested in money, perhaps this could have
accounted for the age effect in younger subjects .-.choosing
fewer Sas. However, this investigator is not aware of any
evidence that supports the hypothesis that six-year-olds are
less interested in money than are ten-year-olds. Further
more, Longstreth (1971 ) has reported data indicating that
six-year-olds and ten-year-olds do not differ significantly
in their interest for money. In one situation, with a small
number of Ss, he found that the six-year-olds learned at
japproximately the same rate as the ten-year-olds to choose a
cup containing a nickel, rather than a marble.
In a second situation, a class of six-year-old Ss was
asked which they preferred— a marble or a nickel and it was
found that there was an unanimous preference for the nickel.
Although the results of these studies cannot be considered
definitive, they do indicate that six-year-olds are inter
ested in receiving money, and in the first case, there is an
indication that six-year-olds and ten-year-olds have approx
imately equal preferences for money. Thus, there is some
evidence to support the hypothesis that six-year-old Ss
chose fewer Sas, not because of a lower motivation for the
money, but rather because they were not able to infer that
Sa could be instrumental in obtaining a prize.
It is interesting to note the similarity of the results
of this study and Longstreth1s study, in spite of the fact
that there were several changes in procedure. Most likely
then, his findings are best accounted for by the cognitive
explanation. If partial reinforcement, familiarity, or
sequence effects were responsible for his Ss learning Sa,
then Sa would not have been learned in the present study
with these confounding variables removed.
Another point worth noting is that in both Longstreth's
and the present study, the young group of Ss was comprised
of 6-year-olds. Although it has been found in some investi
gations that children as young as 4 years 2 months show
development of inferential processes (Braine, 19 59), it
l
iwould appear from these two studies that inferential
processes develop much later. This author would be more
inclined to agree with Piaget that it is not until around
the age of 7 that the child becomes capable of making logi
cal inferences. Six-year-old Ss are in a stage of transi
tion and for a rather complex problem such as the one this
study presented, it is not surprising to find that they did
not show evidence of inferential behavior.
Applying the findings of this study to everyday life,
we can explain how certain acquired reinforcers such as
i
^social approval are effective in controlling behavior some
of the time but not at other times. Take, as an example,
the social approval of "good" which children receive at
various times from different adults. Is telling the child
"good" always effective in controlling behavior? Most
likely it is not. As an illustration, suppose a baby sitter
is taking care of the child one evening. Also suppose that
the baby sitter has already been given instructions by the
parents as to when the child must go to bed and the child is
aware of these instructions. Then compare this situation to
one in which the baby sitter is given control over the
child's bedtime and the child is aware that his staying up
late will depend on the baby sitter being pleased with his
behavior. Most likely, utterances of "good" from the baby
sitter in the first case would be ineffective as compared
^ __ . 56;
jwith those in the second case. Reasoning from the cognitive
i
point of view, we would say that the child in the first
situation perceived that receiving social approval would
have no relevance to his goal of staying up late, whereas in
the second situation he perceived that it would.
Admittedly the cognitive interpretation presented in
this paper does not answer the question of how the goal of
staying late was acquired in the first place. However, the
purpose of this paper has been to ask if originally neutral
objects can become goals in either of two ways: by being
paired with rewards or by being instrumental in obtaining
rewards. To test these two positions, pre-existing goals
Lave been assumed and it has been found that a cognitive
explanation appears to be better able to account for the way
in which new goals are acquired than a secondary reinforce
ment explanation. It is beyond the scope of this paper to
attempt to demonstrate the ontogenesis of all goals of human
oehavior; however, the question of how goals become acquired
Ln the first place is an important and fruitful area for
future research.
r
h
i
|
i
1
A P P E N D U M
57
Before the present experiment was conducted, another
study had been proposed to test the cognitive vs. the Sr
I
positions. The main hypothesis was that responsiveness to
social reinforcement from powerful and weak evaluators
should differ between young Ss and old £s. More specific
ally, older Ss should be sensitive to the differences of the
evaluators and should respond more to the powerful evaluat
or. Younger Ss, on the other hand, would not be expected to
imake the inference of future rewards and privileges from a
j
'powerful experimenter and consequently, social approval from
either a weak or a powerful experimenter should be equally
effective.
Pilot work was carried out to test this hypothesis.
Thirty-three third graders were used as subjects. Pilot
work revealed considerable difficulty in manipulating the
power of the experimenter as a variable. More specifically,
problems were encountered in attempting to establish the
experimenter as being weak; that is, without power.
The first method used to establish the E as weak con
sisted of having the E mediate the prizes that another adult
(A) owned, i.e., A gave E the prizes to give to S. It was
hoped that A, but not E, would be seen as the sole owner and
controller of rewards. To further reduce E1s power and
increase A's, A left before £s were tested and as he left he
locked up the prizes in a suitcase, took the key and told E
not to give out any prizes while he was gone. Questioning
the Ss after they were tested showed that the power manipu
lation was ineffective. Ss in the Weak Experimenter (WE)
condition still believed E could give them prizes after A
had left.
To further reduce E's power, the procedure was changed
so that E was no longer a mediator of prizes in the WE con
dition. A_ gave out prizes directly to the S while E watched
on but did not handle the rewards. A left before the Ss
were tested, took his rewards with him, and told the E and
the £ ! that he would not be back. Thus, it was assumed that
A would be seen as powerful since he was the only one to
|
’ dispense rewards, and E as weak since he had nothing to do
I " " "
( with rewards. However, questionnaire data revealed that Ss
j
in the WE condition still expected to get rewards during the
test phase of the experiment.
The conclusion from the data of these pilot Ss is that
it is very difficult, if not impossible to establish an
adult as being weak in power within the limitations of the
present experiment. The reasons are fairly obvious when the
social milieu of a student in an elementary school is con
sidered. First of all, adults simply by virtue of being
adults and not children are seen as high status persons.
Second, even though an adult is supposedly left without
tangible rewards there is nothing to prevent the child from
believing that the adult can go and get some more. Finally,
tangible rewards are not the only ways that adults have
power over children, particularly in schools. The children
may have believed, for example, that even without tangible
rewards, the weak experimenter could have informed the
teacher of the child's performance. This belief would have
been reinforced by the child's past history of being taken
out of the class and tested. For all of these reasons, it
is not surprising that the experimental manipulations of the
power variable were not effective. Because of the difficul
ties encountered using power as a variable, this study was
abandoned and a second study was proposed which did not
involve such problems.
I
1
R EFERENCES
61
REFERENCES
Bailey, D. A. A cognitive interpreation of social approval.
Paper presented at the 50th annual meeting of the West
ern Psychological Association, Los Angeles, April 15-
18, 1970.
Black, A. H. & Thomas, L. D. Differential effectiveness of
primary, secondary and social reinforcement in discrim
ination learning of mongoloids. Preliminary note.
Perceptual and Motor Skills, 1966, !23, 585-586.
Braine, M. D. S. The ontogeny of certain logical opera
tions: Piaget's formulation examined by nonverbal
methods. Psychological Monographs, 19 59, 73^, Whole
No. 475.
Bolles, R. C. Theory of motivation. New York: Harper &
Row. New York, 1967.
Bugelski, R. R. The psychology of learning. New York:
Henry Holt, 1956.
Coffer, C. N. & Appley, M. H. Motivation: Theory and
research. New York: Wiley & Sons, 19 64.
Donaldson, T. E. Secondary reinforcement vs. other stimulus
effects in extinction and learning of a new response.
Unpublished doctoral dissertation, Purdue University,
1961.
Ellis, N. R. & Pryer, M. W. Primary vs. secondary rein
forcement in simple discrimination learning of mental
defectives. Psychological Reports, 1958, £, 67-70.
Estes, R. K. Number of training trials and variations in
stimulus cues as factors effecting extinction in chil
dren. Unpublished master's thesis, University of
Southern California, 19 60.
Greenwald, A. G. Nuttin's neglected critique of the law of
effect. Psychological Bulletin, 1966, 6_5, 199-205.
Hall, J. Secondary reinforcement or frustration. Unpub-
6 2 _
lished doctoral dissertation, University of Southern
California, 1964.
jHays, W. Statistics for psychologists. New York: Holt,
Rinehart and Wilson, TS'ETI
Hertel, R. K. Secondary reinforcement as a function of
punishment in children. Unpublished M.A. thesis,
University of Southern California, 1966.
Hilgard, E. R. , Atkinson, R. C. & Atkinson, R. L. Introduc
tion to psychology. (5th ed.) New York: Harcourt
Brace Jovanovicn, Inc., 1971.
Hill, W. F. Sources of evaluative reinforcement. Psycho
logical Bulletin, 1968, 69_, 132-146.
Hull, C. L. Principles of behavior. New York: Appleton-
Century-crotts , inc.", Iy4 J.
Kagan, J. & Havemann, E. Psychology: an introduction. New
York: Harcourt, Brace & World, Inc., iy68.
Kass, N., Wilson, H., & Sidowski, J. B. Effects of number
of training trials upon the development of a secondary
reinforcer with children. American Psychologist, 1964,
19, 451 (Abstract).
Kendler, H. H. & Kendler, T. S. Vertical and horizontal
processes in problem solving. Psychological Review,
1962, 69, 1-16.
Kimble, G. A. Conditioning and learning. New York:
Appleton, 1961".
Leiman, A. H., Myers, J. L. & Myers, N. A. Secondary rein
forcement in a discrimination problem with children.
Child Development, 1961, 32^, 349-353.
Longstreth, L. E. Incentive stimuli as determinants of
instrumental response strength in children. Journal of
Comparative and Physiological Psychology, 19677 55,
39'5-4 01.---------------------------------- —
Longstreth, L. E. Frustration effects rather than Sr
effects in children. Psychonomic Science, 1966a, 4,
425-426.
Longstreth, L. E. Frustration and secondary reinforcement
concepts as applied to human instrumental conditioning
and extinction. Psychological monographs, 1966b, 80,
_____ (Whole no. 11) . ____________________________________________
[
Longstreth, L. E. Secondary reinforcement and frustration.
Paper presented at biennial meetings of the Society for
Research in Child Development, New York, 19 67.
Longstreth, L. E. Personal document, 1969.
Longstreth, L. E. A cognitive interpretation of secondary
reinforcement. A paper presented at the Nebraska Sym
posium on Motivation, 19 70a.
Longstreth, L. E. Tests of the law of effect using open and
closed tasks. Journal of Experimental Psychology,
1970b, 84, 53-5TT----------------------------------
Longstreth, L. E. & Hertel, R. A Test of a cognitive inter
pretation of Sr effects. Unpublished manuscript,
1970c.
Longstreth, L. E. Personal communication from class pro
ject, 1970d.
Longstreth, L. E. Personal communication, 19 71.
Lundgren, J. The effect of social evaluations upon behav
ior: a comparison of secondary reinforcement and cog
nitive interpretations. Unpublished Ph.D. disserta
tion. University of Southern California, 1970.
McNemmar, Q. Psychological Statistics. (3rd ed.) New
York: Wiley, 1962.
Myers, J. L. Secondary reinforcement: A review of recent
experimentation. Psychological Bulletin, 19 58, 55,
284-301.
Myers, N. A. Extinction following partial and continuous
primary and secondary reinforcement. Journal of
Experimental Psychology, 1960, 60^, 172-1/9.
Myers, N. A., Craig, G. J. & Myers, J. L. Secondary rein
forcement as a function of the number of reinforced
trials. Child Development, 1961, 32_, 765-772.
Nunnally, J. C., Duchnowski, A. J. & Knott, P. D. Associa
tion of neutral objects with rewards: Effects of
massed vs. distributed practice, delay of testing age
and sex. Journal of Experimental Child Psychology,
1967, 5, 157-T5T.
Parker, R. K. & Nunnally, j. C. Association of neutral
objects with rewards: Effects of reward schedules on
reward expectancy, verbal evaluation, and selective
65
attention. Journal of Experimental Child Psychology,
1966, 3, 324-312''.
Piaget, J. Judgment and reasoning in the child. London:
Routledge & Kegan Paul Ltd., 1928'.
Sidowski, J. B., Kass, N. & Wilson, H. Cue and secondary
reinforcement effects with children. Journal of Expe-
rimental Psychology, 1965, 69_, 340-342.
Skinner, B. F. Science and human behavior. New York:
MacMillan, 1953.
Weir, M. W. Developmental changes in problem-solving
strategies. In Endler, N. S., Boulter, L. R. &
Osser, H. (Eds.) Contemporary issues in developmental
psychology. New York: Holt, Rinehart and Winston,
Inc., 1968.
White, S. H. Evidence for a hierarchical arrangement of
learning processes. In Lipsitt, L. P. & Spiker, C. C.
(Eds.) Advances in child development and behavior.
New York! Academic Press, 19 65, pp. 187-216.
Winer, B. J. Statistical principles in experimental design.
New York: McGraw-Hill, Inc., 1962.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Autonomic and cognitive indices of semantic conditioning and generalization

PDF

A Preliminary Study Of The Dimensions Of Future Time Perspective

PDF

Instrumental Escape Behavior Applied To Remedial Reading

PDF

Modality and serial position effects in immediate and delayed recall

PDF

Inhibitory-excitatory interaction of conditioned autonomic responses through interoceptive and exteroceptive conditioning

PDF

Verbal And Motor Mediation In Normal Children Of Two Ages And Retarded Individuals

PDF

Long-Term And Short-Term Effects Of Training In Feature Scanning On The Cognitive Tempo Of Fourth-Grade Children

PDF

Dominance And Contiguity As Interactive Determinants Of Autonomic Conditioning

PDF

Eye Contact As An Indicator Of Infant Social Development

PDF

Visual Perception Of Static And Dynamic Two-Dimensional Objects: A Cross-Sectional Study

PDF

Reward Expectancy Strength As Related To The Magnitude Of Frustration In Children

PDF

Reversal And Nonreversal Shift Performance Of Retardates Under Various Motivational And Stimulus Conditions

PDF

Human Performance As A Function Of The Joint Effects Of Drive And Incentive Motivation

PDF

Color Vision Deficiency In Third And Sixth Grade Boys In Association To Academic Achievement And Descriptive Behavioral Patterns

PDF

An Evaluation Of Three Compensatory Education Kindergarten Programs

PDF

Positive And Negative Verbal Reinforcement In The Conditioning Of The Verbal Behavior Of Adolescent Underachieving Delinquents With Dependent, Neutral, And Aggressive Stimuli

PDF

Recognition Accuracy In The Method Of Single Stimuli: A Test Of An Operational Definition Of The Distinctiveness Of Stimuli

PDF

Frustration Versus Secondary Reinforcement In Children Under Continuous And Partial Primary Reinforcement

PDF

Personality factors in reading disability: a psychodynamic approach

PDF

The Visual Fusion Threshold (Vft) Test As A Measure Of Perceptual Efficiency In Kindergarten And First Grade, And As A Possible Predictor Of Later Reading Retardation

Asset Metadata

Creator Bailey, Darena Ann (author)

Core Title A Cognitive Interpretation Of Acquired Reinforcement Effects

Degree Doctor of Philosophy

Degree Program Psychology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,psychology, general

Format dissertations (aat)

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Longstreth, Langdon E. (committee chair), Madigan, Stephen (committee member), Marks, Merle B. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-526977

Unique identifier UC11363229

Identifier 7200536.pdf (filename),usctheses-c18-526977 (legacy record id)

Legacy Identifier 7200536

Dmrecord 526977

Document Type Dissertation

Format dissertations (aat)

Rights Bailey, Darena Ann

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