l
l

3'

l
M

3

3‘

3
l

' I

i
33
In

‘33

3
l

‘1 3
33

, {33'
:3 3

3"
.y
l
‘3‘

—

 

5 , ,

EFFFFE3S 0F 33‘333F33T3VE3’13LUF333‘BF“F33 0F
3‘F333 F‘O F..F3~:3FNF F!‘ PQ““3T3F33 3 333) F‘X'FEF '3L3 F33“. SF
1333 33333333333332! 3.33 333. EF‘ "3335. 333 CHFU‘ 33F3‘3

‘ .
.fi
3“ 3
991.4

3'11

T338333 303‘ 3329. Degree 03 N3. 33
33333331" 33333 STAFF 3333M 333333“!
ANDREW RI‘sFPH GILPFN
333?}.

WWWWM K,

71.,

677 0725

\ \

m

\

\

7

m:

W

LIBRARY

\ A
.‘.L .. .
’ o
.
C
v... v

 

[pulpit 91).. .

IchIgan State

M

 

 

 

m“ ‘
mach“: M?-
i O 7 3:

 

 

ABSTRACT
EFFECTS OF INCENTIVE VALUE AND PERCENTAGE OF

REINFORCEMENT ON ACQUISITION AND EXTINCTION OF AN
INSTRUMENTAL RESPONSE IN CHILDREN

By
Andrew Ralph Gilpin

Forty kindergarten children were each assigned to one
of four treatment groups formed by simultaneously manipu-
lating incentive value (high. i.e. toy. vs. low, i.e. no
toy) and reinforcement schedule (random 50 percent, 1.9. PRF,
vs. 100 percent. i.e. CRF). The children first pulled a
lever to receive marble rewards for 60 trials. and then made
30 additional responses during which no rewards were given.
Latency and duration of each response were noted.

Results suggested that boys and girls reacted dif-
ferently: in general. boys responded faster. However, the
sex variable interacted with other variables. The partial
reinforcement acquisition effect (PRAE) was observed, but
not the partial reinforcement extinction effect (PRE).
Girls who received the high incentive treatment had longer
latencies than girls who received the low incentive treat-
ment. and than boys in either incentive condition. but
only during the acquisition phase. Latencies of high in-
centive §s (regardless of sex) who had experienced CRF did
not differ from those of low incentive‘fis who had exper-
ienced PRF at the beginning of the extinction phase, but
by its end, the former had longer latencies. While there

were no overall sex differences in movement duration at the

Andrew Ralph Gilpin
beginning of the extinction phase, by the end of the phase
girls had longer durations than boys.

Methodological implications for future research were

discussed. Theoretical interpretation emphasized the util-

ity of cognitive dissonance theory.

Acce% (3 ,
Date 5: 4763/

 

EFFECTS OF INCENTIVE VALUE AND PERCENTAGE OF
REINFORCEMENT ON ACQUISITION AND EXTINCTION OF AN
INSTRUMENTAL RESPONSE IN CHILDREN
BY

Andrew Ralph Gilpin

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF ARTS

Department of Psychology
1971

A."’-7)
. ‘J

DEDICATION
For my mother. who taught me how to write. and
my father. who taught me to read what I had

written.

11

ACKNOWLEDGMENTS

The author expresses his appreciation to the Holt
Public School system. Holt. Michigan, and to the faculty.
staff, and students of the Elliott Elementary School for
their cooperation in carrying out the present study.

The author wishes to express his thanks to Jean Gilpin.
Hark Jackson. Paul Martin, and Steve Plous. who assisted in
data collection.

A special debt of gratitude is owed to the members of
the author's Committee, Dr. Ellen.A. Strommen. Dr. Stanley
C. Ratner. and particularly Dr. Hiram E. Fitzgerald. Chair-
man, for their contributions to the successful completion
of all phases of the project. During the course of the
study's execution. the author served as Research Assistant
to Dr. Ratner. who was kind enough to allow him to concen-
trate on it. at times at the expense of his other duties.

This study was supported in part by a Michigan State
University Research Grant and an NIMH research grant MH-
18655-01 to Hiram E. Fitzgerald. Use of the Michigan
State University computer facilities was made possible
through support. in part. from the National Science Foun-
dation.

111

TABLE OF CONTENTS

LIST OF TABLES. . . . . . . . . . . . .
LIST OF FIGURES. . . . . . . . . . . .
Introduction. . . . . . . . . . . . . .
Method. . . . . . . . . . . . . . . . .
Subjects. . . . . . . . . . . . . . .
Apparatus. . . . . . . . . . . . . .
Procedure. . . . . . . . . . . . . .
Results. . . . . . . . . . . . . .
Acquisition Phase, Latency. . . . . .
Acquisition Phase, Movement Duration.
Extinction Phase, Latency. . . . . .
Extinction Phase, Movement Duration.
Discussion. . . . . . . . . . . . . . .
Summary. . . . . . . . . . . .
LIST OF REFERENCES. . . . . . . . . . .
APPENDIX. . . . . . . . . . . . . . .
Letter to Parents. . . . . . . . . .

1V

LIST OF TABLES

Table 1. Acquisition Latency

Means for Significant Effects. . . . . . .
Table 2. Acquisition Latency Analysis of Variance.
Table 3. Acquisition Movement Duration

Analysis of Variance. . . . . . . . . . . . .
Table A. Extinction Latency

Means for Significant Effects. . . . . . . . .
Table 5. Extinction Latency

Analysis of Variance. . . . . . . . . . . . .
Table 6. Extinction Movement Duration

Analysis of Variance. . . . . . . . . . . . .
Table 7. Extinction Duration

Means for Significant Effects. . . . . . . .
Table 8. Number of Experimenter Comments to

Subject. . . . . . . . . . . . . . .
Table 9. Analysis of Variance

of Experimenter Comments. . . . . . . . . . .

.1h
. 15

. 19
.21

LIST OF FIGURES

Figure 1. Mean acquisition latency as a function of
incentive. schedule. sex. and trials. . . . . . .16
Figure 2. Mean latency as a function
of schedule and trials. . . . . . . . . . . . . .17
Figure 3. Mean movement duration as a fu ction of

Behedule and trials 0 C O O O I O O O O O I O O O 20
Figure A. Mean extinction latency as a function
of schedule, incentive, and trials. . . . . . . .22

Figure 5. Mean extinc ion movement duration
as a function of sex and trials. . . . . . . . . 27

V1

INTRODUCTION

The partial reinforcement acquisition effect (PRAE),
the phenomenon of superior asymptotic performance level of
an instrumental response by subjects (gs) receiving partial
reinforcement schedules to that of gs receiving continuous
reinforcement. has been noted frequently in studies where
children performed a lever pull task (Chertkoff, 1968;
Nakamura, 1969; Pederson. 1970; Ryan. Orton. & Pimm, 1968;
Ryan. Strawbridge. & Matters. 1969; Ryan & Watson. 1968;
Semler & Pederson. 1968). On the other hand. the partial
reinforcement effect (PRE) refers to greater resistance to
extinction in gs reinforced on a random partial reinforce-
ment schedule (or PRF, henceforth assumed to be 50 percent
unless otherwise indicated) during acquisition compared with
those §s who received 100 percent or continuous reinforce-
ment (CRF).

The PRE has been successfully demonstrated with a va-
riety of infrahuman species (Bitterman & Schoel. 1970); but
attempts to demonstrate the PRE in children with the lever
apparatus have often been unsuccessful. when the dependent
variable was latency. duration, or amplitude of response
(Bruning. 1964; Holton, 1961; Rosenbaum & Bruning. 1966;
Ryan & Voorhoeve. 1966. These variables are not all highly

correlated: cf. Cantor & Kubose. 1970). However. as Ryan
1

2
and Voorhoeve (1966) pointed out. when rate of lever pull-
ing (responses per unit time) or trials until the child
quit responding have been measured. the PRE has been ob-
served (Baumeister & Hawkins. 1966; Bijou. 1957; Chert-
koff. 1968; Fattu. Mech. & Auble. 1955; Kass. 1962; Kass &
Wilson. 1966; Lewis. 1960; Nakamura. 1969). One study. re-
ported by Chertkoff (1968). found that regardless of wheth-
er §s were required to continue responding. or were free to
quit whenever they wished. movement speeds did not reveal
the PRE; but more §s who had received CRF during acquisi-
tion quit before completing 30 extinction trials than did
§s who had received PRF.

Thus. while it is clear that the PRAE can be demon-
strated in children using lever pull speed as a measure of
performance. the status of the PRE in this context is less
clearcut.

Incentive value is another variable which has been
found to influence children's performance, particularly in
probability and discrimination learning tasks. where in-
vestigators have frequently found that increased incentive
level facilitates performance (e.g.. Bisett & Rieber. 1966;
Brackbill & Jack. 1958; Offenbach, 1964). although in some
cases. low incentive groups did better (Miller & Estes.
1961; Stevenson & Hoving. 1964; Stevenson & Weir. 1959).
Efforts to extend such findings to a lever apparatus. us-
ing marbles as rewards. have been generally unsuccessful.
Ryan (1965) asked kindergarten children to rank six toys

prior to a series of lever pulls. He told half of them

3
that if they ”won the game" they would receive their most-
preferred toy; he told the other children that if they won.
they would receive their least-preferred toy. In addition.
half of each group received a PRF schedule; the other half
received CRF. Although Ryan did find evidence of the PRAE.
incentive value did not influence response level.

Ryan and Moffitt (1966) performed a study similar to
Ryan's. but made three modifications. First. only one lever
was involved (Ryan. 1965. used two levers). Second. in ad-
dition to kindergarteners. preschool Se were included.
Third. the incentives were a ten-cent toy (high) and a piece
of string (low). Ryan and Moffitt found that for kinder-
garteners. there was no Schedule X Incentive interaction for
starting (reaction) speeds. For preschoolers. the PRF low
incentive group performed faster than the PRF high incentive
group; there were no differences between CRF groups. With
respect to movement speeds (speed of response completion
once begun). there was no significant Schedule X Incentive
interaction for either age. but low incentive §s increased
speed over trials faster than high incentive §s (cf. Watson.
1967).

Sheikh (1968) manipulated goal distance at which delay
of reward was introduced when kindergarten children pulled
a lever. He used two incentive conditions: high (preferred
toy) vs. low (a balloon). and found no differences in eith-
er starting speed or movement speed as a function of incen-

tive 0

None of the above studies reported significant main

1.

effects for the incentive manipulation. Possible explan-
ations of the failure to obtain reliable differences due
to varied incentive value are that differences between low
and high incentive conditions were not great enough. or
that the task itself. which was invariably presented as a
game. was intrinsically so interesting that it masked the
effect of incentive differences (Ryan. 1965; Sheikh. 1968).
The latter possibility is supported by studies successfully
demonstrating the PRAE or the PRE where marbles could not
be redeemed for a toy at the end of the session (Chertkoff.
1968; Ryan. gt g;.. 1968; Ryan.‘§§'§;.. 1969). as well as
by the positive results obtained in learning tasks.

An issue closely related to incentive value is the
function served by marble rewards. The early experiments
in the lever pull task frequently provided that marbles
could be redeemed at the end of the experiment for a prize
(Ryan & Watson. 1968). and when this is the case. marbles
are true token reinforcers. However. in at least one
study. children kept the marbles they had won (Chertkoff.
1968). In other instances. §s could only see the marbles
(e.g.. Nakamura & Ellis. 1964; Ryan. 33 al.. 1968; Ryan. 25
§;.. 1969). In these instances it is possible that the mar-
bles served only an information function. i.e. they indi-
cated to §.that he was playing the game well (despite the
fact that reward contingencies were independent of response
level). The distinction is particularly important in view
of Watson's (1970) speculation that gs may be either "rein-

forcer-oriented" or ”solution-oriented."

5

Since in some sense §s who participated in experiments
where prizes were given presumably experienced higher in-
centive motivation than §s who participated in studies where
no prizes were given. some experimenters have implicitly
used a high incentive condition. and others have used a low
incentive condition. All have made the tacit assumption
that whether or not marbles were tokens was independent of
such phenomena as the Frustration Effect. the PRAE. and the
PRE. This assumption is consistent with failure to find
differences due to incentive value. as well as with studies
which did not use marble rewards but nonetheless demon-
strated the PRAE (e.g.. Bruning. 1964; Ryan & Watson. 1966;
Watson. 1967). But it is difficult to reconcile with Ryan
and Moffitt's (1966) finding of an Incentive X Schedule in-
teraction in preschoolers. as well as with Terrell's (1958)
work with discrimination learning and Walters and Foote's
(1962) work with probability learning: in both of the latter
instances. performance was better for children who received
token reinforcers than for children who received rewards
that were merely informative.

While incentive value manipulations have been attempted
within the context of the Frustration Effect (Ryan. 1965).
the PRAE (Ryan & Moffitt. 1966). and goal distance (Sheikh.
1968). the possible interaction between incentive value and
reinforcement schedule as they affect resistance to extinc-
tion has not received empirical attention gig g’gig lever
pull studies with children. Equally important. since the

situation where there is no prize at the end of the

6

experiment constitutes the most extreme instance of a low
incentive condition. its inclusion in the present study
provides the most rigorous possible test of the theory that
previous studies did not include a sufficiently wide range
in incentive value.

The present study was designed to accomplish several
ends: (1) to provide a more sufficient manipulation of incen-
tive value than previous research; (2) to extend study of
the incentive variable to an extinction phase; (3) to pro-
vide additional evidence with regard to the issue of whether
the PRE can be observed using response speed as a dependent
variable. under conditions where §,is required to continue
responding; and (4) to resolve the methodological issue of
whether or not allowing Se to exchange marbles for a prize
affects the PRAE. the PRE. or both.

Method
Subjects

Forty kindergarten children (Mean CA=5 years 6 mo..
SD=5 mo.). 23 boys and 17 girls. were each assigned to one
of four groups according to incentive value (high vs. low)
and schedule (CRF vs. PRF). Ten Se were assigned to each
of the groups. Assignment to schedule condition was ran-
dom. In order to prevent low incentive §s from being a-
ware that some other children.were winning prizes, all Se
in each of two classes were assigned to the same incentive
value. Subjects from a third class were randomly as-
signed to the two conditions. but all those designated as

low incentive gs were tested before their high incentive

classmates.1

Apparatus

The apparatus has been described elsewhere (Davidson &
Fitzgerald. 1970). In the present study. the second and
third levers of the Davidson and Fitzgerald triple lever
console were removed and their slots covered flush with the
surface.

The basic unit was a console approximately the height
of an elementary school desk. The lever was mounted on the
front panel. which measured 22 in. by 30 in.. The lever .
moved from right to left along a 16 in. slot. located to-
ward the rear of the console; it protruded approximately
six inches above the surface of the console. and had a
plastic handle. A small green arrow located directly in
front of the initial position of the lever acted as a cue
for lever movement. The S was seated on a chair in front
of the console. the height of which could be adjusted so
that the lever was in easy reach.

When the arrow lighted up (it was controlled by a
switch on the control unit. described below). signalling
the start of a trial. a timer was started automatically.
This timer. which measured latency of response. was con-
tained in a control unit connected to the console by a

cable; one‘g operated this unit and recorded data. Initial

 

V

1The project was approved by the school board. the princi-
pal. and teachers. Parents were sent a letter explaining
the study; only children who returned permission slips
were used as S8. Each child was asked if he wished to
participate prior to the session; two did not and were not
run as Se. Data sheets were coded so that anonymity was
ensured.

8
movement of the lever activated a magnetic reed switch.
which stopped the latency timer and started a second
timer (movement duration). When the lever was moved to
the end of its slot. this second timer was automatically
stopped. Hence measures of latency and movement times were
available for each response.

Mounted on the left side (as §_faced the apparatus) of
the console was a plastic tube down which a second EImanual-
ly dispensed marbles on rewarded trials. The tube ter-
minated in a small metal cup. The second E’sat behind the
console. facing S; when E dropped a marble down the tube his
hand was hidden from S6 view by a shield. The child took
the marble and deposited it in a goal box located on his
right. The goal box was essentially a painted cigar box
with a hole in the top; it prevented §,from examining the
accumulated marbles between trials. and controlled for the
goal distance factor sometimes found in such studies (Hol-
ton. 1961; Pederson & McEwan. 1970. Ryan & Voorhoeve.
1966. suggested the use of such a box.).

Other materials included clear glass marbles of dif-
ferent colors. a variety of inexpensive toys such as small
cars and modelling clay. a stopwatch for timing intertrial
interval. a counter used by‘g to record trial number. and
a set of pictures of Disney cartoon characters.

Procedure

Subjects were tested individually in a room provided

by the school. The E briefly interacted with gland then

gave one of the following directions:

9

(Low Incentive Condition) "Do you like to play games.
.____7 Well. you can play our game here today. But first I
have some pictures I'd like to show you. Maybe you recog-
nize some of them -- is there anyone you know? Which is
your favorite? (The child chooses from the cartoon pictures)
I like that. too. But now let's play the game. To play
you sit with your hands in your lap and watch this arrow.
As soon as it lights up (demonstration). you move this le-
ver all the way over to the end of the slot. Do you think
you can do that? When you move the lever. sometimes you'll
get a marble -- it'll roll down this tube into the cup
here. When you get one. pick it up and drop it into this
box. If you can get enough marbles. then you'll win the
game.”

(High Incentive Condition) "Do you like to play games.
____3 Well. you can play our game here today. But first I
have some toys I'd like to show you. If you can win the
game. you may keep one of these toys. Which is your favor-
ite? (The child chooses) All right. We'll put that over
here (next to goal box). and if you win. you may keep it.
But now let's play the game. To play. you sit with your
hands in your lap and watch this arrow. As soon as it
lights up (demonstration), you move this lever all the way
over to the end of the slot. Do you think you can do that?
When you move the lever, sometimes you'll get a marble ~-
it'll roll down this tube into this box. If you can get
enough marbles. then you'll win the game and you can keep

the toy.”

10

Low incentive §s looked at the pictures to control for
social interaction experienced by high incentive §s while
they chose their toys. The E'who interacted with §.and
administered marble rewards was always male (there were four
such Es ) .

All §s were then asked. "Do you understand? We'll try
a couple of practice times first, and then start to play.
Remember. watch the arrow. and as soon as it lights up. move
the lever. Ready?" The "Ready?” was actually a cue to the
second E|running the controls. telling him to turn on the
arrow light. After two rewarded trials (more if §_seemed
confused). the acquisition phase began. All §s received 60
acquisition trials. spaced eight to ten seconds apart
starting from completion of lever movement. After rewarded
trials. S deposited the marble in the goal box during this
period; E’returned the lever to its initial position. On-
set of each trial was signalled with "Ready?"; if §,was not
orienting to the arrow. "Watch the arrow!" was added.

For the PRF condition. the 60 trials were composed of
15 blocks of four trials each. with two rewards in each
block; order within blocks was random. All Se received the
same schedule: RNRN.RRNN.RNRN.RNNR.RNRN.NRNR.RRNN.NRNR.NNRR.
RNRN.NNRR.RRNN.NNRR.NNRR.RRNN. All CRF gs received a marble
after each of the 60 responses.

After Sireceived the 60 trials of the acquisition phase.
the extinction phase began. All §s received 30 extinction
trials. with no rewards administered. On the last trial.

the response resulted in delivery of a final marble. Then

11

§,told all gs they had won the game. High incentive §s
received their toys. The SS were accompanied back to their
classrooms.

After the session had begun. E responded to Ss' com-
ments only by saying "That's the way the machine works;”
no other comments were made by E. Because of the potential
social reinforcement involved (of. Ryan & Watson. 1966). a

record was kept of all such comments.2

When all experi-
mental §s had been run. toys were given to children who had

not already received them.

Results

Because of the possibility of contradictory conclu-
sions arising from the use of speed versus time measures in
a skewed distribution (Edington. 1960). prior to analysis a
logarithmic transformation was performed on raw data. which
were expressed in units of hundredths of a second. In an
effort to reduce some of the error variance. the analysis
treated sex as an independent variable. Because in three
of the four Incentive X Schedule cells there were six boys
and four girls. data from boys were discarded (along with
that of one girl) randomly so that four boys and four girls
remained in each cell (i.e.. total N = 32). It was felt
that since this approach necessitated a rather complex
analysis. and since only the beginning and end trials of the

two phases. acquisition and extinction. were really of

 

2A one-way Analysis of Variance (Winer. 1962. p. 55) of the

numbers of comments in each of the eight Incentive X Sched-

u e X Sex cells failed to reveal significant differences
(7'24)‘1056g p,010].

12
relevance in testing the hypotheses of interest. data
from intervening trials in each phase could be omitted
without invalidating the results (Obviously. this pro-
cedure precluded any examination of the shape of perfor-
mance curves. e.g. with respect to linear vs. quadratic
trend.).

Hence the means of trials 1 through 4; 57 through 60;
61 through 64; and 87 through 90 -- i.e.. the first and
last four trials of acquisition and extinction -- were
treated as the ”raw" data. prior to transformation. Sepa-
rate analyses were performed for the two phases. and laten-
cy and movement times were treated separately. The Analysis
of Variance design used in all four instances was an ex-
tension of a Lindquist (1953. pp. 281-284) Type III mixed
design. with repeated measures on Factor C (Trials). The
design involved Between-S effects for A (Incentive Value).

B (Schedule). D (Sex). AXB. AXD. BXD. and AXBXD; Within-S
effects were C (Trials). AXC. BXC. CXD. AXBXC. AXCXD.
BXCXD. and AXBXCXD (2x2x2x2 levels).

All means reported are in units of Log10(seconds x 100%
Thus. for example. the mean latency value over all Se and
trials was 2.101. and the mean for duration was 1.716; their
antilogs are 1.262 seconds and 0.520 seconds respectively.
Acguisition.§ha§g. Latency

Table 1 and Figures 1 and 2 present the mean latency
scores; the analysis appears in Table 2. As can be seen.
there was a significant sex effect: boys responded faster

than girls. There was also a significant Incentive X Sex

13
interaction. A posteriori comparisons by use of the New-
man-Keuls procedure (Winer. 1962. pp. 80-85; 210) revealed
that girls in the high incentive condition had signifi-
cantly (p<.01) longer latencies than girls in the low in-
centive condition. or boys in either incentive condition;
the latter three groups did not differ significantly from
one another. The main effect for trials was significant:
latencies were shorter at the end of the acquisition phase
than at the beginning. There was a quadruple interaction
between incentive. schedule. trials. and sex. appearing in
Figure 1. Most of the differences represented were not
reliable. Analysis by means of the Newman-Keuls procedure
revealed that the only significant differences occurred
among certain high incentive groups: PRF males at the end
of acquisition had significantly shorter latencies than
CRF females at the beginning of acquisition (p<.01). the
PRF females at the beginning of acquisition (p<305). and
the PRF girls at the end of acquisition (p<.05). No other
differences were significant. In addition. there was a
trend toward shorter latencies for low incentive gs com-
pared with high incentive §s. but it failed to reach statis-
tical significance.

Because there was substantial g_priori evidence pre-
dicting the PRAE. several planned comparisons were performed
on the Schedule X Trials interaction. despite the low sta-
tistical significance level for the effect itself (For a
defense of this technique. see Winer. 1962. p. 85). The

data appear in Figure 2. There was no significant difference

14

Table 1. Acquisition Latency Means for Significant Effects.

Level(s) Mean In Log(sec's x 100)

 

Effect: D(Sex)

 

 

 

Males 2.080
Females 2.145
Effect: AXD(Incentive X Sex)

Low. Males 2.085
Low. Females 2.100
High. Males (2.076
High. Females ”2.190

 

 

Effectz‘C(Trials)

Trials 1-4 2.133
Trials 57-60 2.093

Effect: A(Incentive)

Low 2.092
High 2.133

Note.--See Figures 1 and 2 also.

15
Table 2. Acquisition Latency Analysis of Variance.

 

 

 

 

 

 

 

 

 

Source df MS F

Between Se 31
A(Incentive) 1 0.02646832 3.364a
B(Schedule) 1 0.00256676 0.326
D(Sex) 1 0.06715022 8.534**
AXB i 0.00063693 0.081
AXD 1 0.03830930 4.868*
BXD i 0.00789463 1.003
AXBXD 1 0.00208131 0.264
Error 24 0.00786883
Within‘gs 32
C(Trials) 1 0.02588495 7.011*
AXC l 0.00023289 0.063
BXC 1 0.00499422 1.353
CXD 1 0.00136490 0.370
AXBXC i 0.00015912 0.043
AXCXD 1 0.00142479 0.386
BXCXD 1 0.00465070 1.260
AXBXCXD 1 0.02059552 5.578*
Error 24 0.00369202

a.05<p<.08

*p<.05

**p<.01

16

«45¢... «30“. “.0 9.00.5
m.$m_u...o_m mhmm+nm_

 

P r P
.. 2:- 5
u u | llfu II. “Xhm.
my... Lu Emu... I .3333

:2...- Dfoauoxg. .. .62..

 

 

 

00 .N

EON

.m..N

.ON.N

 

(001x Spas m
ADNBIV‘I) 90': man:

.mamahp use .Nmm .oasnonom .obapSmona ho noduonsu m we monopma nofipamasdom use: .H ehswam

17

.mawanp use eazomnom no scauossm m we monopma use: .m ohsmam

35¢... «50“. mo mxuOJm

 

mu .N m. t m. n. : m N m m _
. . oo.N
20:02.th “ 20..._._m_300< W
_ . 3
_ . W0 NINI W
3."... . m m
“an. _ I . ores. w
“EU _ H n
_ w m
_ . TS: w
_ A
_
. . cud.

 

18

between CRF and PRF §s at the beginning of acquisition
[F(1.24)=1.9§]; there was no significant difference between
the two groups at the end of acquisition [F(i.24)=0.543:
there was no significant difference in the interaction de-
fined by magnitude of the decrease in latency over the
phase for the two schedule groups [F(1.24)=1.3§l. That is.
the PRAE was not obtained using latency as the dependent
variable.
Acquisition Phage. Movement Duration

The analysis appears in Table 3. The only effect ap-
proaching significance was the Schedule X Trials inter-
action. represented in Figure 3. Again. planned compari-
sons were performed. While schedule groups did not differ
significantly at the beginning of acquisition [F(1.24)=
0.30]. by trials 57-60 CRF gs had longer movement durations
than PRF _S_s [F(1.24)=6.62. p<.05]. and the interaction de-
fined above and indicated in Figure 3 was significant [§(1.
24)=4.95. p<.0§]. The PRAE was obtained using this measure.
Extinction Phage. Latency

Latency data appear in Table 4 and Figures 2 and 4; the
analysis appears in Table 5. As was true for the acqui-
sition phase. there was a trend toward shorter latencies
among Se in the low incentive condition than for those in
the high incentive group. but it fell short of statistical
significance.

Again. while the expected overall Schedule X Trials
interaction was not observed planned comparisons were

performed. There was no significant difference between PRF

Table 3. Acquisition Movement Duration Analysis of Variance.

~wh-I‘

19

 

 

 

Source df MS F
Between Se 31
A(Incentive) 1 0.07131194 2.226
B(Schedule) 1 0.03182726 0.994
D(Sex) 1 0.05669062 1.770
AXB 1 0.00342583 0.107
AXD 1 0.0159169? 0.497
BXD 1 0.00015366 0.005
AXBXD 1 0.04355784 1.360
Error 24 0.03202889
Within‘Ss 32
C(Trials) 1 0.02666921 1.724
AXC 1 0.03139146 2.023
BXC 1 0.07560439 4.888*
CXD 1 0.00591433 0.382
AXBXC 1 0.00861612 0.55?
AXCXD 1 0.0058008? 0.375
BXCXD 1 0.01939288 1.254
AXBXCXD 1 0.00231386 0.150
Error 24 0.01546849

 

 

 

 

*p<.05

20

.mamahp use mazoosom mo Godpoczm m we Scapsaho pzoaobos use: .m masmdm

351... So“. no 960.6

 

 

 

nu .u m. t m. n. : a s m m..
— T s p p " tome.
zo_._.oz:xm _ zocasgc .
_ .06..
_
«35 "
kc: AK.—
mm _
.00..
:00;

_
_
.
_
_

(00: x spas m ami
iNBNBAOm 901 NVBIN

 

21

Table 4. Extinction Latency Means for Significant Effects.

 

 

Level(s) Mean [In Log(sec's x 1003
Effect: A(Incentive)

Low 2.063
High 2.124

Effect: AXBXC(Incentive X Schedule X Trials)

Low, CRF. Trials 61-64 2.103
Low. CRF. Trials 87-90 - 2.058
Low. PRF. Trials 61-64 2.021
Low. PRF. Trials 87-90 2.070
High. CRF. Trials 61-64 2.088
High. CRF. Trials 87-90 2.170
High. PRF. Trials 61-64 2.100
High. PRF. Trials 87-90 2.139

 

Effect: BXCXD(Schedule X Trials X Sex)

 

CRF. Trials 61-64. Males 2.115
CRF. Trials 87-90. Males 2.079
CRF. Trials 61-64. Females 2.076
CRF. Trials 87-90. Females 2.148
PRF. Trials 61-64. Males 2.016
PRF. Trials 87-90. Males 2.078
PRF. Trials 61-64. Females 2.104
PRF. Trials 87-90, Females 2.130

 

Note.-- See Figures 2 and 4 also.

22

.mamaap one .obapcoosa .oasoosom ho coauocsm a we honouma sodpocdpxo secs .3 ohsmam

mast... $30“. no mxuOJm

 

nN NN .N ON 0. m. h. 0..

uOOd
\\.
315 \\ \\ .\ .\ \\ \\ e m N
Row-.3339 \.\\\\ .mo N .I
“.53.... .3? \ m. m
S .1
i... 1... II . I O_.N x «VI
\\\\\\ I I.
“.239... :e...D..\.\\.\.\ W m
t m ~0N ( A

.1393... :2:
.ON..N

 

23

Table 5. Extinction Latency Analysis of Variance.

 

 

 

 

 

 

 

ource df MS F

Between S8 31
A(Incentive) 1 0.05989738 3.078a
B(Schedule) 1 0.00807361 0.415
D(Sex) 1 0.02895994 1.488
AXB 1 0.00265048 0.136
AXD 1 0.02898389 1.489
BXD 1 0.0120055} 0.617
AXBXD 1 0.00980676 0.504
Error 24 0.01946085
within as 32
C(Trials) 1 0.01546142 2.909
AAC 1 0.0138210} 2.600
BXC 1 0.00262565 0.494
CXD 1 0.00506895 0.954b
AXBXC 1 0.01902879 3.580
AXCXD 1 0.00727563 1.369o
BXCXD 1 0.02094406 3.940
AXBXCXD 1 0.01476914 2.779
Error 24 0.00531510

a.05<p<.10

b.05<p<.08

°.OS<P<.06

24
and CRF §s at the beginning of the extinction phase [F(1.
24)=1.87]. or at the end of the phase [F(1.24)=0.1§]; nor
was the interaction defined above significant [F(1.24)=
0.49]. That is. the PRE was not observed.

Two higher order interactions involving the schedule
and trials variables were also suggested by the analysis.
viz. an Incentive X Schedule X Trials interaction. and a
Schedule X Trials X Sex interaction. Since in neither case
did the significance level exceed p=.05. the performance of
a posteriori comparisons must be considered theoretically
tenuous.

With regard to the Incentive X Schedule X Trials inter-
action (see Figure 4). the only significant difference [New-
man-Keuls, p<.0£| was between high incentive. CRF §s at the
end of the extinction phase and low incentive. PRF §s at
the beginning of the phase: the latter performed faster.
Upon examination. the Newman-Keuls analysis of the Schedule
X Trials X Sex interaction failed to detect any differences
that were significant [at p<.01; the difference between CRF
girls at the end of the phase and PRF boys at the beginning
was significant at p<.05. but should probably be discounted
under the circumstanceé].

Extinction Phase. Movement Duration

As indicated in Tables 6 and 7. there was a significant
trials effect; §s took longer to perform the response at the
end of the phase than at its beginning. Although the Sched-
ule X Trials interaction was not significant. the relevant

planned comparisons were nonetheless performed. At the

ET?“— TTf-f‘jfih F.

25

Table 6. Extinction Movement Duration Analysis of Variance.

 

 

 

 

 

 

 

 

 

Source df MS F

Between.§s 31
A(Incentive) 1 0.00827719 0.344
B(Schedule) 1 0.04626373 1.921
D(Sex) 1 0.02323168 0.965
AXE 1 0.01465444 0.608
AXD 1 0.05469780 2.271
BXD 1 0.00458385 0.1903
AXBXD 1 0.081060u7 3.366
Error 24 0.02408298
Within _S_s 32
C(Trials) 1 0.04936328 5.435*
AXC 1 0.00216980 0.239
BXC 1 0.00534268 0.588
CXD 1 0.04756499 5.237*
AXBXC 1 0.00008242 0.009
AXCXD 1 0.00625628 0.689
BXCXD 1 0.01762162 1.940
AXBXCXD 1 0.00217910 0.240
Error 24 0.00908316

3. 05<p<. 08

*p<.05

.a u .u nun-.1; em: . any
I

26

Table 7. Extinction Duration Means for Significant Effects

 

 

 

 

Level(s) [ Mean InELogIsec's x 100”
Effect: C(Trials)

Trials 61-64 1.678

Trials 87-90 1.733

 

Effect: CXD(Trials X Sex)

 

 

 

 

Trials 61-64. Males 1.686
Trials 61-64. Females 1.669
Trials 87-90. Males 1.687
Trials 87-90. Females . 1.780
Effect: AXBXC(Incentive X Schedule X Sex)
Low. CRF. Males 1.773
Low. CRF. Females 1.699
Low. PRF. Males 1.635
Low. PRF. Females 1.669
High. CRF. Males 1.636
High. CRF. Females 1.821
High. PRF. Males 1.701
High. PRF. Females 1.709

 

 

Note.-- See Figures 3 and 5 also.

27

 

.mamaap one new we scaponsm a me sodumaso pcmsmbos godponauao use: .m magmam

m...<.m.k (30.... .....O mxuoqm

 

 

nu «N .N ON m. w. t m.
.00..
$0..
«Wicw
.flu4<15'\
5N4
:3...
«04(89-
r00..

 

(00: x 9935 NI 3qu
iNBNI-Mom 901 NVBW

28

beginning of extinction. PRF Se performed significantly
faster (shorter duration) than CRF §s [F(1.24)=4.58. p<’
.05; see Figure 3]. This presumably represents a contin-
uation of the performance levels at the end of the acqui-
sition phase. By the end of extinction. there was no
significant difference between schedule groups [F(1.24)=
1.1T]; also. the interaction as defined above was not
significant EF(1.24)=0.59]. 3

There was a significant Trials X Sex interaction (see 5
Figure 5). which was unanticipated. A Newman-Keuls analysis 8
revealed that at the end of extinction. girls were taking
longer than they did at the beginning of that phase. or than
boys at either beginning or end of extinction [all signifi-
cant at .01<p<;05]; there were no other significant dif-
ferences. There was also a trend toward an Incentive X
Schedule X Sex interaction. but this was not significant.
A Newman-Keuls analysis failed to find any differences sig-
nificant at p<.05.

Discussion

The major purposes of the present study were (1) to
demonstrate the PRAE and the PRE. (2) to determine whether
incentive value pp; pg affects response speed. and (3) to
evaluate any interaction between the schedule and incentive
variables. uriefly. the results indicated that (1) the
PRAE was successfully shown. but the PRE was not observed;
(2) while evidence was inconclusive. incentive value did
affect response speed for some Se. and (3) the schedule and

incentive variables did interact. More specifically. girls

29
who received the high incentive treatment had longer
latencies than girls who received the low incentive treat-
ment, and than boys in either incentive condition. but only
during the acquisition phase. Latencies of high incentive
§s (regardless of sex) who had experienced CRF did not dif-
fer from those of low incentive §S who had experienced PRF
at the beginning of the extinction phase. but by its end.
the former had longer latencies. While there were no over-
all sex differences in movement duration at the beginning
of the extinction phase. by the end of the phase girls had
longer response durations than boys.

In general. the status of sex differences in lever
pull performance has been ambiguous. Ryan and Voorhoeve
(1966) found no sex differences in a parametric study of
the PRAE and PRE with kindergarteners. Ryan and Watson
(1966) reported similar findings using social reinforce-
ment instead of marble tokens. Indeed. in reviewing the
frustrative nonreward literature dealing with children.
Ryan and Watson (1968) omitted sex effects from their dis-
cussion of individual differences that had been found to
influence lever pull behavior. Interestingly enough. none
of the previous researchers manipulating incentive value
with children pulling levers (viz. Ryan. 1965; Ryan & Mof-
fitt. 1966; Sheikh. 1968) bothered to report possible in-
teractions with sex. despite the fact that all used S3 of
both sexes.

Nevertheless. recent evidence suggests that the lack

of a sex effect may be confined to children less than six

30
years of age. Ryan. Orton. and Pimm (1968). in an in-
vestigation of developmental changes in instrumental lever
pulling behavior. employed gs ranging from kindergarten
age to 75 years. and found that at all levels except kin-
dergarteners. movement speed was faster for males than for
females. The results obtained in the present study sug-
gest that the sex effect may also be present for kinder-
garteners. when latency of response is measured. This in-
terpretation is supported by comments made by several boys.
who expressed the desire to pull so fast "the lever will
break."

Within the lever pull context. there is no report of
a differential sex effect on the PRAE or the PRE in child-
ren (sex may interact with magnitude of reward and inter-
response interval; cf. Nakamura. 1969; Nakamura & Ellis.
1964; Pederson. 1970). The results of the present study
suggest that future researchers should look more closely at
the sex variable. since there were a variety of inter-
actions between sex and both reinforcement schedule and in-
centive value.

Nakamura (1966). who had children manipulate a joystidh
found an interaction between magnitude of reward (relative
number of marbles per reward) and sex as they affected lever
activity (rate of response) during extinction (of. Nakamura
& Ellis. 1964). While the relation of this finding to the
present study is somewhat unclear. the theoretical model
which Nakamura used to explain his results may be relevant.

He interpreted his findings in terms of cognitive dissonance,

. , ,"‘.—-' T'TT” “—‘- r
F _
F ‘ .

31
based on previous research (Nakamura & Ellis. 1964) which
indicated that girls were more persistent than boys (re-
sponded longer during extinction). following high magni-
tude of reward.

Lawrence and Festinger (1962) applied cognitive dis-
sonance considerations. derived from personality research.
to rats running alleys under partial reinforcement condi-
tions. The essence of their position, as they stated it.
is that "If an organism continues to engage in an activity
while possessing information that. considered alone. would
lead it to discontinue the activity. it will develop some
extra attractions for the activity. or its consequences. in
order to give itself additional justification for contin-
uing to engage in the behavior.”(Lawrence & Festinger. 1962.
p. 156). These extra attractions involve ”drives and moti-
vations" which are subordinate to the drive which is prompt-
ing the response; extra attractions refer more to competing
motives than to competing responses. though the latter are
presumably involved somehow. Thus. while a rat runs an al-
ley primarily to obtain food. it could also satisfy other
motives than hunger. perhaps the most obvious of which is a
need for stimulus variation. Similarly, while a child may
pull a lever to win a marble. he might also do so for other
reasons. for instance. to please §,or because playing the
game is inherently pleasurable (rather than frustrating).

A child who experiences CRF, however. will not develop
dissonance or extra attractions under Lawrence and Festinge‘s

model. Receiving a marble after each response is consistent

32

with the cognition that one is working hard pulling the
lever. In contrast. a S who experiences a PRF schedule
does develop dissonance. If he wishes to win the game.
he must amass many marbles. and the only way to do this
is to continue responding. When he eXperiences a non-
rewarded trial. he realizes that (1) he has just expended
effort. and (2) he has no marble to show for it. These
cognitions are mutually contradictory and hence §,ex-
periences dissonance. One can reduce dissonance best by
omitting the behavior in question. but as pointed out. this
will not win the game. But if S can find other reasons to
continue responding. his dissonance will also be reduced.

Lawrence and Festinger predicted that asymptotic acqui-
sition performance should be higher for CRF §s than for PRF
‘Ss. This is consistent with the then (1962) available
studies with animals. but stands in opposition to the PRAE
observed with children. Bitterman and Schoel (1970) sug-
gest that the PRAE occurs under massed trials. with relative-
ly large numbers of trials. and this finding may provide a
means of extending dissonance theory. Lawrence and Festin-
ger assumed that speed of response increases as a function
of the number of rewarded trials during acquisition. If
this effect is supposed to reach an asymptote after some
number of massed rewarded trials. then eventually a PRF
group would "catch up" to the CRF group. Extra attractions
might then explain the PRAE. At any rate, this extension
does not alter the dissonance explanations of extinction.

Lawrence and Festinger were chiefly concerned with the

33
PRE; they suggested that PRF gs would be able to maintain
high performance levels during extinction because of the
availability of extra attractions. CRF §s would es-
sentially start "from scratch" in that they would ex-
perience dissonance for the first time during extinction
(the same is true for frustration under Amsel's 1958. 1962
theory). Moreover. continued responding would not result
in rewards. whereas it did in the case of PRF §s during
acquisition. The best way to reduce dissonance for CRF Se
is to stop responding: while §.won't win the game this way.
at least no effort is expended. Thus PRF §s should show
more resistance to extinction than CRF §S.

While the present study does not provide a means of
pitting cognitive dissonance theory against its chief ri-
vals. sequential learning theory (Capaldi. 1966. 1967.
1970) and frustration theory (Amsel. 1958. 1962), in any
systematic way. it does present certain findings amenable
to a dissonance interpretation. One important finding
emerging from Nakamura's (1966) work is that the relative
value (in terms of its exchange value) of a marble reward
increases when expected rewards are not forthcoming (similar
findings are reported by Knott. Nunnally. & Duchnowski.
1967. and Lewis. 1964; they follow directly from the posi-
tion that extra attractions are develOping). While it is
difficult to predict how this phenomenon would affect in-
centive value. it does illustrate the fact that an effec-
tive manipulation of the latter may not be reflected in

the reinforcing value of a particular marble reward.

1“ MEIJA' ‘

34
According to Spence's (1956) theory. both the reinforcing
value of the marbles and overall incentive value should
have the same effect on motivation. Insofar as the incen-
tive value manipulation of the present study was actually
represented in Ss' performance. the trend toward faster
speeds for low incentive gs (significant for girls during
the acquisition phase) seems to cast doubt on the veridi-
cality of that theory. But_while this finding is difficult
to explain within a frustration framework. dissonance
theory does predict it.

One might postulate that low incentive §s experience
some dissonance regardless of their reinforcement schedule:
they are exerting effort for comparatively small payoff.
at least compared with high incentive Ss. The latter can
justify their effort with the cognition that continued re-
sponding can result in their obtaining a toy. Extra at-
tractions could develop in the low incentive group. re-
sulting in their responding faster relative to high incen-
tive SS. This explanation is also in accord with the
findings of Ryan and Moffitt (1966) that whereas movement
speeds of high incentive gs remained stable during acqui-
sition, those of low incentive §s increased as a function
of trials.

Ryan and Moffitt explained their finding in terms of an
anger hypothesis. viz. that low incentive §s were angered
when. having rated the string gig a,p;§ the toy. they were
required to play for the former. thus operating under in-

creased drive level. Sheikh's (1968) low incentive gs were

35

unaware of the existence of more desirable incentives. and
did not respond faster than high incentive Se; Ryan and
Moffitt claimed this as evidence for their contention.
However. Sheikh's low incentive prize was a balloon. prob-
ably more desirable than a piece of string. At any rate.
the anger hypothesis is rendered untenable by the present
study. Low incentive §s were unaware of the possibility of
winning toys. and yet responded either at the same speed .1
or faster than high incentive §s. 3

One way to test the dissonance interpretation of in- :-
centive value effects would be to replicate Ryan and Mof-
fitt's study. but in addition asking S3 to rank the string
and the toy after performance of the lever pull task. If
the low incentive‘Ss developed extra attractions. one might
expect that the string would become relatively more desir-
able (the "sour grapes" phenomenon might depress the value
of the toy. but this would also follow from dissonance
theory.).

With regard to extinction. one might expect from a dis-
sonance position that both variables -- schedule and incen-
tive -- would operate so that the most resistance to ex-
tinction should occur for low incentive gs who received PRF
during acquisition. and the least for high incentive §s who
received CRF. This is precisely what was observed in the
Schedule X Incentive X Trials interaction for extinction
latencies. Ryan (1965) made the prediction that nonrein-
forcement would be more frustrating under high incentive

than under low incentive treatments. The frustration

36

prediction would then be that response speed would be
highest for a high incentive. PRF group and lowest for a
low incentive. CRF group. and presumably this relation
would continue during extinction. Clearly this prediction
derives little support from either Ryan and Moffitt's
study or the present investigation.

Another phenomenon which the dissonance model predicts
is the finding that the PRE occurs where S responds volun- 3.
tarily (e.g.. Chertkoff. 1968; Kass. 1962; Kass & Wilson. g
1966). The present study is in accord with previous work ;1
using speed of response. where S had to continue responding
whether he wanted to or not, in that there was no clearcut
evidence of the PRE. Ryan and Voorhoeve (1966). in con-
sidering this problem. pointed out that two circumstances
in their study could have constituted constraints on ex-
tinction. First. marbles acquired during acquisition were
on display during extinction and might have distracted‘S.
This explanation is ruled out by the present study. where
accumulated marbles were not visible to S. Second. complete
extinction might have been prevented by S's instructions to
S that he (S) was to pull the lever every time the stimulus
light came on. This is merely a variation of the hypothesis
that extinction is enhanced when responding is voluntary.
and it seems than enough evidence exists to tentatively ac-
cept it as veridical. While it does not actually contra-
dict frustration theory. it can be derived from the frustra-
tion position only if one assumes that frustration is en-

hanced when S has the opportunity to avoid frustration but

37
does not choose to do so.

However. a variety of evidence derived from research
on personality theory suggests that dissonance is increased
when Sh dissonance-arousing behavior is voluntarily emit-
ted (Shafer. 1968); hence. development of extra attractions
might be facilitated. and a greater PRE observed. (Indeed.
some dissonance theorists would insist that volition is
necessary for dissonance to develop. for example. Brehm &
Cohen. 1962. The present author disagrees, as does Kiesler.
1968.)

The findings of interactions between incentive value
and the other variables suggest that it may be inappro-
priate to equate results found in studies where marbles
served as token reinforcers with those of studies in which
no toy prize was involved. That is. the PRAE and PRE may
exist only under particular incentive conditions. and if
so, token reward may be an important variable.

Most previous experiments with children pulling levers
under varied schedules of reinforcement have found that
movement speeds or times were more stable measures of re-
sponsivity than latency, presumably due to the effects of
competing responses on the latter (Cantor & Kubose. 1970;
Ryan & Cantor. 1962). The present study suggests that this
may be true only when the child is free to ignore the
"ready" light. When the trial is not begun until S is
orienting to the light. more reliable results are obtained.
However. latencies did not reveal the PRAE. Moreover. the
procedure necessitated S's sitting facing S. and the re-

sulting possibility of a confound with social reinforcement

38
prooably outweighs the advantages of this technique.
Since the lever in the present study was not spring-loaded.
§,had to return it after each lever pull, which was another
reason for his position; however. this could easily be cor-
rected (as has generally been done in such studies).
The observed increase in movement duration. inter-
preted as extinc ion. may have been due in part to the ef-

fects of fatigue. The kindergarten §s made a total of 90 ?1

53-.-

responses spaced only eight to ten seconds apart. which

-..- ow

would seem likely to tax the endurance of children of that
age. rhe fatigue interpretation cannot be firmly dis-
counted in the absence of control groups which continued
to receive acquisition reward contingencies over the 30
trial extinction phase. and any extensions of the study
should employ such controls. However. there is some evi-
dence suggesting that the effects of fatigue may not have
been so great. First. Ryan and Voorhoeve (1966). using
such control groups. found no differences between them
and the extinction groups except toward the end of the

30 extinction trials; if fatigue were a confounding vari-
able. such differences should have disappeared. Un—
fortunately. Ryan and Voorhoeve's §s made only 70 total
responses. Using slightly older gs (6- and 7-year-olds).
Kass and Wilson (1966) found that many gs who were re-
sponding voluntarily made as many as 240 responses be-
fore quitting. Regrettably. previous work (Kass. 196‘)
indicates that younger §s (4-year-olds) make far fewer

voluntary responses than the older group. However. if §s

39
in the present stndy were becoming fatigued. it is likely
that latencies would have shown an equal or greater in-
crease than did movement duration. yet there was no indi-
cation of this.

In short, while fatigue may have been a factor for
some gs (and may have contributed to interactions involving
sex during extinction. if girls were more susceptible to
it). it is unreasonable to attribute failure to obtain the

PRE to its effects. All §s made the same number of re-

‘g‘m ...-_..——_.-— n ‘L-‘
_ i r I

sponses. 1
Summary

The results of the present study suggest that future
researchers in this area must pay attention to the sex of
their §s as a possible confounding variable. They must also
consider carefully whether or not to use a toy as an in-
centive; cognitive dissonance theory may be useful in re-
gard to this issue. which is rather peripheral to the frus-
tration theory typically applied in such studies. Low in—
centive gs sometimes respond as though they were under high-
er motivational level than "high" incentive gs.

While the veridicality of the PRAE in this context is
well established. the use of a procedure where §s respond
voluntarily seems advisable in studies involving the PRE.
This also seems well explained by dissonance consider-

ations.

LIS‘i‘ OF REFERENCES

LIST OF REFERENCES

Amsel. A. The role of frustrative nonreward in noncontin-
uous reward situations. Psychological Bulletin. 1958.
fie 102-1190 ‘

Amsel. A. Frustrative nonreward in partial reinforcement
and discrimination learning: Some recent history and ;
a theoretical extension. Psychological Review. 1962. ij
ég, 306-328. :.

Baumeister. A. A.. & Hawkins. w. F. Extinction and disin-
hibition as a function of reinforcement schedUIe with
severely retarded children. {gurnal of Experimental
Child Psychology. 1966.-2; 3h3-3h7.

Bijou. S. w. Patterns of reinforcement and resistance to
extinction in young children. Child Development.
19579 §§. “7‘5“.

Bisett. B. M.. & Rieber. N. The effects of age and incen—
tive value on discrimination learning. Journal of
Experimental Child Psychology. 1966._2. 199-206.

Bitterman. M. E.. & Schoel. w. M. Instrumental learning in
animals: Parameters of reinforcement. énnual Review
of PsychologyJ 1970._g;, 367-“36. 'fi'

 

Brackbill. 1.. & Jack. D. Discrimination learning in child-
ren as a function of reinforcement value. Child Devel-
opment. 1958. 32, 185-190.

 

Brehm. J. N.. & Cohen. A. R. Explorations in cognitive dis-
sonance. New York: diley. 1962.

Bruning. J. L. Effects of magnitude of reward and percentage
of reinforcement on a lever movement response. Child
Development, 196h. 35. 281-285.

Cantor. G. N.. & Kubose. S. K. The correlation between
amplitude and speed measures in children's lever-pul-
ling behavior. Psychonomic Science. i970..l§. 83-84.

Capaldi. E. J. Partial reinforcement: A hypothesis of se-
uential effects. Psychological Review. 1966..22! 459-
77.
no

41

Capaldi. E. J. A sequential hypothesis of instrumental
learning. In K. N. Spence & J. T. Spence (Eds.).
The psychology of learning and motivation. Advances
In researCh and theory, vol. I. pp. 67-156. New
York: Academic Press. 1967.

Capaldi. E. J. An analysis of the role of reward and
reward magnitude in instrumental learning. In J. H.
Reynierse (Ed.). Qurrent issues in animal learning.
A colloquium. pp. 357-389. Lincoln. Nebraska: Univ.
of Nebraska Press. 1970.

 

Chertkoff. J. M. Response speed under voluntary and invol-
untary conditions as a function of percentage of rein—
forcement. Child Development. 1968..29. 167—175.

Davidson. N. H.. & Fitzgerald, H. E. Recency and summation
effects of nonreward in children. Journal of Experi-
mental Child Psychology. 1970. 19. 16-27.

 

r—._.__ *_.5_ TV

 

Edington, E. S. Contradictory conclusions from two speed of
performance measures. _Esychological Bulletin, 1960,

:19 315‘3170

Fattu. N. A.. Mech. E. V.. & Auble. D. Partial reinforce-
ment related to "free" responding in extinction with
preschool children. {gurnal of Experimental Education.

1955. 22. 365-368.

Heber. R. F. Motor task performance of high grade mentally
retarded males as a function of magnitude of incentive.
gmerican Journal of Mental Deficiency. 1959. 63, 667-

71. “‘

Holton. R. B.. Amplitude of an instrumental response fol-
lowing the cessation of reward. Child DevelOpment.
1961’ 18—, 107-1160

 

Horowitz. F. D. Incentive value of social stimuli for pre-
school children. Child Development. 1962,_22, 111-116.

 

Kass. N. Resistance to extinction as a function of age and
schedules of reinforcement. Journal of Experimental
EsychOIOSYo 1962._é&} 249-252.

 

Kass. N.. & Wilson. H. Resistance to extinction as a function
of percentage of reinforcement. number of training
trials. and conditioned reinforcement. JQurnal of
Experimental Psychology. 1966. 21. 355-357.

 

 

Kiesler. C. A. Commitment. Chapt. 37 in Abelson. Aronson.
McGuire. Newcomb. Rosenberg. & Tannenbaum (Eds.).
Theories of cognitive consistency: A sourcebook.
Chicago: Rand McNally. 1968.

42

Knott. P. D.. Nunnally. J. C.. & Duchnowski. A. J. Ef-
fects of frustration on primary and conditioned in-
centive value. Journal of Experimental Research in
Personalityj 1967. g, 140-1h9.

Lawrence. D. R.. & Festinger. L.. Deterrents and rein-
forcement: The psychology of insufficient reward.
Stanford: Stanford Univ. Press. 1962.

 

Lewis. D. J. Partial reinforcement: A selective review of
the literature since 1950. Psychological Bulletin.
1960' 2' 1-280

 

Lewis. M. Effect of effort on value: An exploratory study
of children. Child Development, 196k. 25. 1337-1342.

 

Lindquist. E. F. Design and analysis of experiments in
psychology and education. Boston: Houghton Mifflin.
1953.

Miller. L. B.. & Estes. 3. W. Monetary reward and motivation
in discrimination learning. Journal of Experimental
Psychology, l961._§1. 501-504.

 

 

 

 

 

Nakamura. C. Y. Effect of prominence of dissonance—associa-
ted stimuli during evaluation of the stimuli. Journal-
of Experimental Child Psychology, 1966._3, 86-99.

Nakamura. C. Y. Effects of increasing and decreasing reward
magnitude and pre-experimental persistence level on fo—
cal and incidental responses. Journal of Experimental
Child Psychology. 1969,_z, sin-531.

Nakamura. C. Y.. & Ellis. F. F. Methodological study of the
effects of relative reward magnitude on performance.
Child Development, 1964, 35, 595-610.

Offenbach, S. I. Studies of children's probability learning
behavior: I. Effect of reward and punishment at two
age levels. Child Development. 196h._25, 709-715.

 

Pederson. D. R. Children's reactions to failure as a func-
tion of interresponse interval. Research Bulletin
No. 171. December. 1970. Department of Psychology.
University of Western Ontario. London. Canada.

Pederson, D. R.. & McEwan. R. C. Children's reactions to
failure as a function of instructions and goal dis-
tance. Journal of Experimental Child Psychology. 1970.
29 51-58e

Rosenbaum. M. E.. & druning. J. L. Direct and vicarious
experience of variations in percentage of reinforcement

Child Development, 1966..}]. 959-966.

43

Ryan. T. J. The effects of nonreinforcement and incentive
value on response speed. Child Development, 1965._2§.
1067-1082.

Ryan. T. J.. & Cantor. G. N. Response speed in children as
a function of reinforcement schedule. Child Develop-
aaui. 1962.11. 871-878.

 

Ryan, T. J.. & Moffitt, A. R. Response speed as a function
of age, incentive value. and reinforcement schedule.
Child Development. 1966, 27. 103-113.

 

Ryan. T. J.. Orton. C.. & Pimm, J. B. Discrete-trial instru-
mental performance related to reward schedule and
developmental level. Journal of Experimental Psychol-
251’ 19689 780 31‘370

Ryan. T. J.. Strawbridge, J. E.. & Watters. R. G. Patterning
in partial reward schedules with children. Paper pre-
sented at the Biennial Meeting of the Society for Re-
search in Child Development. Santa Monica. Calif.. 1969.

Ryan. T. J.. & Voorhoeve, A. C. A parametric investigation
of reinforcement schedule and sex of S as related to
acquisition and extinction of an instrumental response.
Jgurnal of Experimental Child PsychOIOSY. 1966.;&.

1 9.197.

Ryan. T. J.. & Watson. P. Children's response speeds as a
function of sex and verbal reinforcement schedule.
Psychonomic Science. 1966._§J 271-372.

Ryan. T. J.. & Watson. P. Frustrative nonreward theory ap-
plied to children's behavior. Psychological Bulletin.
1968. 62. 111-125.

 

Semler, I. J.. & Pederson, D. R. Children's reactions to
nonreward: Partial vs. continuous reinforcement using
a within-subjects design. Psychonomic Science. 1968.
19. 285-286.

Shafer. J. The effect of nonreward dissonance and secondary
reward on extinction and attractiveness. (Doctoral dis-
sertation. University of Southern California) Ann Ar-
bor. Mich.: University Microfilms. 1968. No. 68-13.587.

Sheikh. A. A. Children's response speed as a function of
delay of reward at different distances from the goal
and incentive value. Psychonomic Science. 1968. 11.
201.

Spence. K..W. Behavior theory and conditioning. New Haven:
Yale Univ. Press. 1956.

 

40

Stevenson. H. W.. & Roving. K. L.. Probability learning
as a function of age and incentive. Journal of
Experimental Child Psychology. 196h._1. 64-70.

Stevenson. H. W.. & Weir. M. W. Variables affecting
children's performance in a probability learning

task. qurnal of Experimental PsychologY. 1959.
27, hOBIHIZ.

Terrell. G. The role of incentive in discrimination
learnigg in children. Child DevelOpmentJ 1958. 29.
231-33 .

Walters. R. H.. & Foote, A. A. A study of reinforcer ef-
fectiveness with children. Merrill-Palmer Quarterly.
1962 . ._8+ 1119-157.

 

 

Watson. P. Effects of reward schedule. type of reward and
isolation condition on children's lever-pulling per-
formance. Unpublished Doctoral dissertation. Univer-
sity of Western Ontario. 1967.

Watson. P. Individual differences in children's reactions
to frustrative nonreward. Journal of Experimental
Child Psychology. 1970._ig, 216-23U.

 

Winer. B. J. Statistical principles in experimental design.
New York: McGraw-Hill. 1962.

APPENDIX

APPENDIX

 

 

 

Table 8. Number of Experimenter Comments to Subject.

 

4

 

 

 

 

 

Incentive Schedule Sex Mean No. of Comments Mdn. No.
Low CRF M 1.00 0.00
Low PRF M 1.00 0.50
High CRF M 1.25 0.00
High PRF M 0.75 0.50
Low CRF F 0.75 0.50
Low PRF F 0.75 0.00
High CRF F 1.00 0.00
High PRF F 1.25 1.00

 

Note.--n=h|§s per cell.

45

#6

Table 9. Analysis of Variance of Experimenter Comments

‘—

 

 

_§ource df MS F p
Condition 7 n.3o 1.52 >.1o

 

 

 

 

Error 2h 2.82

 

1:7
Letter 22 Parents
A copy of the letter sent to parents of potential §s

follows.

MICHIGAN STATE UNIVERSITY. East Lansing 48823

 

Department of Psychology - Olds Hall
Dear Parent:

One of the most pressing questions facing teachers and par-
ents today is how to reward the child when he is performing
some task well. Must the child be rewarded after every cor-
rect response. or is this only necessary part of the time?
What kind of reward works best; is the knowledge that he is
doing well sufficient. or must the child be given some sort
of tangible incentive such as a new toy?

We are trying to answer these questions by examining types
of rewards and frequencies they are given. We would like
very much for you to allow your child to participate in this
study with the other children in his class.

Each child devotes about half an hour, during which time he
plays a game in which he pulls a lever to receive marbles;
at the end of the session, he can exchange the marbles for
a small toy. He will take part in only one session. and the
study will not detract from his work in the classroom. This
doeslggt involve any sort of personality or intelligence
test; our pilot work has shown that children really enjoy
participation in the experiment. The game fascinates them
and the toy is of course welcome.

The Holt School District Superintendent and the School Princi-
pal have already given their approval to the project. Please
use the form below to indicate whether or not you consent to
your child's being included in the study (Your child cannot

be included unless the form is returned to the teacher). If
you need more information please contact Mr. Gilpin (tel, f)
weekdays in the evening.

We appreciate your Prompt consideration for cooperation in
this study. On request, an interpretive summary of the re-
sults will be sent to you at completion of the project.

Sincerely.

Dr. Hiram E. Fitzgerald
Mr. Andrew R. Gilpin
Please check one of the following alternatives:
___I wish my child to be included in the study described.
___I doing; wish my child to be included in the study.
Parent
Date
Child's Name
(PLEASE RETURN THIS COMPLETED FORM AS SOON AS POSSIBLE}

 

 

 

48

"MllllillllAll“