w-vwvvvmwuw—n“‘I‘I‘Q‘l‘fi',’1‘wmfl"3‘9II'Yii'Yfi“
' ‘ ‘ ."‘.—V .n3u".'.‘. ‘7',‘ .4. '

REINFORCEMENT RELAY Roms EFREcIs ON
BERARQRAL CONTRAST " """

   

' The 13 1.051113 Degree Rf Fh D
ARCLAGAR STATE LARRERSLTY """
RALRR WILLAM RICHARDS x, _, I
' ’ 1971 A . '  * 7'; 131i;

 

LIB E’ARY

l\\\\3\\\\\\1\\2\\1§\\\\1\\ All ALMA! “531$?“

 

This is to certify that the

thesis entitled

REINFORCEMENT DELAY:
SOME EFFECTS ON BEHAVIGIAL CONTRAST

presented by

Ralph William Richards

has been accepted towards fulfillment
of the requirements for

__Eh.D.._degree in Elf-1191931

 

AW

0-7639

 

 

’W? 7

 

 

ABSTRACT

REINFORCEMENT DELAY: SOME EFFECTS ON
BEHAVIORAL CONTRAST

BY

Ralph William Richards

Interactions among component schedules of a
multiple schedule of reinforcement were examined. Spe-
cifically. the present study sought to determine if
delaying reinforcement during one component of a mul-
tiple schedule would increase responding during the
other component, i.e.. produce behavioral contrast.

Also examined was the relationship between the duration
of reinforcement delay and the amount of behavioral con-
trast.

The 35 White Carneaux pigeons were conditioned
in one of two standard operant conditioning chambers.
Subjects received 20 sessions of nondelayed reinforce-
ment on a two-component multiple schedule that had iden-
tical variable interval 1-minute schedules as the compo-
nents. The training stimuli were a 555 nm. light and a

l

Ralph William Richards

white vertical line superimposed on a 555 nm. surround.
For the remaining 15 sessions. subjects were assigned to
one of 5 groups. with 7 subjects per group. Four of
these groups received reinforcement according to the
same multiple schedule as before. but with reinforcement
during one component delayed for either 2.5. 5. 10. or
120 seconds. The fifth group was switched to a multiple
variable interval 1-minute extinction schedule of rein-
forcement.

Results showed that the delaying of reinforce-
ment during one component of a multiple schedule does
produce behavioral contrast. Groups that received the
various durations of reinforcement delay or even extinc—
tion during the altered component did not. however. show
a statistically significant difference in the amount of
behavioral contrast. In terms of theoretical implica-
tions. it was suggested that neither a reduction in rein—
forcement frequency or response rate during the altered
component is necessary to the production or maintenance

of behavioral contrast.

Approved: flL I. é (’( '

1"

Date: ‘ j ,_ (I
I
I

REINFORCEMENT DELAY: SOME EFFECTS ON

BEHAVIORAL CONTRAST

BY

Ralph William Richards

A THESIS

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

DOCTOR OF PHILOSOPHY

Department of Psychology

1971

To Nancy

ACKNOWLEDGEMENTS

I would like to express my sincere thanks to

Dr. Mark Rilling whose unrestrictive and yet instructive

gamma-ha gar-r—

supervision and guidance in the laboratory provided the
environment from which this thesis emerged. Dr. Rilling
also served as an excellent advisor and chairman of the
dissertation committee. for which I am also grateful.
Thanks are also due to Drs. M. Ray Denny. Hiram E.
Fitzgerald. and Gordon Wood for their advice and service
as members of the guidance and dissertation committee.
I would also like to thank Jay Wright and
Dr. Glenn I. Hatton for their instruction and assistance
on the use of certain photographic equipment and mate-
rials. Without their help. clear visual presentation of

the results would not have been possible.

iii

LIST OF TABLES . .

LIST OF FIGURES. .

INTRODUCTION . . .

METHOD . . . . . .

RESULTS. . . . . .

DISCUSSION . . . .

LIST OF REFERENCES

TABLE

OF CONTENTS

iv

Page

vi

33

4O

LIST OF TABLES

Table Page

1. RESPONSES PER MINUTE DURING Sl OVER THE SIX
3-DAY BLOCKS (SESSIONS 23-40) FOR THE
GROUPS WHICH RECEIVED EITHER NO REINFORCE-
MENT OR REINFORCEMENT THAT WAS DELAYED FOR
2.5. 5. 10 OR lZO-SEC. DURING $2 . . . . . . 25

2. SUMMARY TABLE OF THE ANALYSIS OF VARIANCE ON
THE 81 RESPONSE RATES OVER THE SIX 3-DAY
BLOCKS (SESSIONS 23-40) FOR THE GROUPS
.WHICH RECEIVED EITHER NO REINFORCEMENT OR
REINFORCEMENT THAT WAS DELAYED FOR 2.5. 5.
10. OR lZO-SEC. DURING 52. . . . . . . . . . 26

3. PEARSON PRODUCT-MOMENT CORRELATIONS BETWEEN
THE MAGNITUDE OF THE INCREASE IN 31 RESPOND-
ING AND THE MAGNITUDE OF THE DECREASE IN
52 RESPONDING. . . . . . . . . . . . . . . . 26

LIST OF FIGURES (Cont.)
Figure Page

4. Response rates to S1 (x's) and $2 (circles)
over sessions. During the 10 sessions to
the left of the dashed vertical line. the
schedule of reinforcement was a multiple
VI l—min. VI l-min. with no reinforcement
delay. During the 15 sessions to the
right of the dashed vertical line. sub-
jects continued to be reinforced on the
same multiple VI l-min. VI l-min. schedule.
but with reinforcement during S2 delayed
for lZO-sec.. . . . . . . . . . . . . . . . 22

5. ReSponse rates to S1 (x's) and $2 (circles)
over sessions. During the 10 sessions to
the left of the dashed vertical line. the
schedule of reinforcement was a multiple
VI l-min. VI l—min. with no reinforcement
delay. During the 15 sessions to the
right of the dashed vertical line. sub—
jects were reinforced on a multiple VI
l-min. extinction schedule. . . . . . . . . 24

6. Reinforcement rates during Sl (x's) and S
(circles) over blocks of sessions (3
sessions per block). During the 3 blocks
of sessions to the left of the dashed
vertical line. the schedule of reinforce—
ment was a multiple VI l-min. VI l-min.
with no reinforcement delay. During the
5 blocks of sessions to the right of the
dashed vertical line. subjects continued
to be reinforced on the same multiple VI
l-min. VI l-min. schedule. but with rein-
forcement during 82 delayed for 2.5-sec.. . 28

2

7. Reinforcement rates during S (x's) and S
(circles) over blocks of sessions (3
sessions per block). During the 3 blocks
of sessions to the left of the dashed
vertical line. the schedule of

2

vii

INTRODUCTION

A multiple schedule requires successive presenta-
tion of two or more independent schedules of reinforcement.
each in the presence of a different exteroceptive stimulus.
Much of the recent research on multiple schedules (e.g..
Reynolds. 1961a; 1961b; 1961c; Reynolds & Limpo. 1968;
Terrace. 1968; Weisman. 1969; 1970) has examined the fre-
quent interactions that occur between the component sched—
ules. An interaction occurs when an organism's behavior
during one component in which reinforcement contingencies
are held constant is in some way affected by a change in

the reinforcement contingencies associated with the other

 

component.

 

Behavioral contrast has been the most widely
studied interaction and has been reviewed extensively by
Terrace (1966a; 1971a) and Dunham (1968). Behavioral con—
trast occurs when the response rate during the unaltered

component increases and moves in a direction away from the

 

 

response rate during the altered component. For example.

1

the changing of a multiple variable interval l-min. var-
iable interval l-min. (multiple VI l-min. VI l-min.)
schedule to a multiple variable interval l-min. extinction
schedule has consistently produced behavioral contrast.
i.e.. the response rate during the first component is
higher when it is alternated with an extinction conponent
than when it is alternated with another variable interval
l-min. component. Behavioral contrast is not. however.
limited to the special multiple schedule that has extinc-
tion as one of its components. as it has been demonstrated
in many other multiple schedules. such as multiple vari-
able interval l-min. variable interval 5-min. (Guttman.
1959; Terrace. 1968; Weisman. 1969). multiple fixed in—
terval l—min. fixed interval 3—min. (Staddon. 1969).
multiple variable interval fixed ratio (Reynolds. 1961c;
Thompson. 1965: Bloomfield. 1967). multiple variable in-
terval differential reinforcement of low response rates
(Bloomfield. 1967; Terrace. 1968; Weisman. 1969). and
multiple variable interval differential reinforcement of
other behavior (Weisman. 1970). It has also been observed
when responses during one of the multiple schedule's com—
ponents are punished with electric shock (Brethower &

Reynolds. 1962; Terrace. 1968).

Recent research has required the rejection or
modification of several theories of behavioral contrast
(e.g.. Reynolds. 1961a; Terrace. 1966a; 1966b; 1968).
For example. the demonstration that behavioral contrast
can occur without a reduction in reinforcement frequency
(Brethower & Reynolds. 1962; Reynolds & Limpo. 1968;
Terrace. 1968; Weisman. 1969; 1970) has led to the re—
jection of Reynolds' relative frequency of reinforcement
hypothesis. Similarly. Terrace (1971a) has modified his
earlier position following Wilkie's (1970) demonstration
that the delivery of reinforcement independently of the
organism's behavior during one component does not produce
behavioral contrast and following the suggestion that
satiation during one component would not produce behav-
ioral contrast. given that different reinforcers were
used in the two components (cf. Premack. 1969). While
Terrace still. apparently. views behavioral contrast as
a by-product of frustration or some other emotional re—
Sponses. these emotional responses are now seen to be
aroused by a reduction in response rate during one com-

ponent that is produced by an inhibition of responding.

 

In each of the previous studies. reinforcement was

delivered immediately upon completion of the schedule's

 

requirements. Following the suggestion of several inves—
tigators (Brown & Farber. 1951; Holder. Marx. Holder. &
Collier. 1957; Amsel. 1958; Renner. 1964) that delayed
reinforcement elicits frustration. the present experiment
examined the effects of delaying reinforcement during one
component of a multiple VI l-min. VI l—min. schedule.
Specifically. the present experiment attempted to deter—
mine if delaying reinforcement during one component would
produce behavioral contrast. Also examined was the rela-
tionship between the duration of reinforcement delay and

the amount of behavioral contrast.

 

METHOD

Subjects

The thirty-five experimentally naive. adult. female
White Carneaux pigeons were maintained at approximately 80%

of their free—feeding weight.

Apparatus

An Industrial Electronics Engineers in-line display
cell (model lO-3723—757-L). mounted behind the right key.
illuminated each of the two Operant conditioning chambers
(Lehigh Valley Electronics. model 1519). A minimum force
of approximately 15-g. was required to operate this key;
the other keys were covered with masking tape. No house-
1ight was used. but a white pilot light was attached to the
rear wall of both chambers.

During reinforcement the key light was extinguished
and a light within the food aperture illuminated. Rein-
forcement was 2.75-sec. access to mixed grain. The timing
of this interval began when the subject placed its head

5

runs”- .

through the food aperture. thus. interrupting a light beam
focused on a photocell.
Standard programming and recording equipment was

housed in an adjacent room.

Procedure

 

Throughout training the key was illuminated by a
555 nm. light (81) or by a white vertical line superimposed
on a 555 nm. surround (82). These stimuli were not ad-
justed for equal intensity. The duration of each S and

l

52 presentation was 30—sec.. excluding reinforcement time
and reinforcement delay time. The stimuli were presented
in random order with the restriction that neither stimulus
occur more than 3 times in succession. A timeout. during
which the chamber was dark and responding nonfunctional.
separated all stimulus presentations. During session 1 the
timeout duration was 1-sec.. while during all subsequent
sessions it was 5—sec. Reinforcements set up but not col-
lected during one component did not carry over to the next.
The pigeons were conditioned to key peck by the

method of successive approximation. The shaping session

terminated following 30 consecutive key pecks. Sessions 2

q A '- 0 .r-nn-rdrnn
..

and 3 also terminated after 30 continuous reinforcements.
During sessions 4 and 5. 3O reinforcements were delivered
according to fixed ratio 10 and 20 schedules. respectively.
All subjects. then. received 20 sessions (60 stimulus pre—
sentations per session) of reinforcement according to a
multiple VI l-min. VI l-min. schedule. During all of the
above sessions. there was no delay of reinforcement.

Each of the subjects was. then. assigned to one of
5 groups. with 7 subjects per group. Four of these groups
received 15 additional sessions of multiple VI l-min. VI
l-min. reinforcement; but with reinforcement during 82 de-
layed for either 2.5. 5. 10. or lZO-sec. During the delay
period. the key light was extinguished and the pilot light
on the Chamber's back wall was illuminated. At the end
of the delay period the pilot light was extinguished and
the food magazine raised. The fifth group received 15
sessions of multiple VI 1-min. extinction training; re-

sponding to S extinguished the key light for lZO-sec.

2
(according to the same VI l-min. schedule). but no rein-

forcement was delivered. All subjects were reinforced for
responding to S1 as before. VI l-min. with no reinforcement

delay. Four of the subjects in each group were run in one

of the conditioning chambers and the remaining subjects

were run in the other chamber.

,H b __ -_.

 

RESULTS

Figures 1-5 show the reSponse rates to S (denoted

1

by the X's) and S (denoted by the closed circles) during

2

the last 25 sessions of training. Each figure represents
one experimental group and contains the data for the 7 in-
dividual subjects as well as the group mean. To the left
of the dashed vertical line are the final 10 sessions
under the multiple VI l—min. VI l-min. schedule with no
delay of reinforcement. During the 15 sessions to the
right of the dashed vertical line the contingencies for S2
reSponding were altered as indicated by the caption above
each figure. That is, the subjects represented in Figures
1—4 continued to be reinforced on the same multiple sched—

ule as before, but with reinforcement during S delayed

2

for either 2.5, 5, 10, or lZO-sec. The subjects in

Figure 5 were not reinforced for reSponding to S during

2

these last 15 sessions.
As shown in Figure l delaying reinforcement for

2.5-sec. during S produced different effects on the S

2 2

9

lO

reSponse rates of the individual subjects. The S2 response
rates were either temporarily increased and then permanently
reduced (bird 38), permanently reduced (birds 609 and 30),

temporarily reduced (birds 199 and 206), or not affected

(birds 2751 and 4304). The group mean evidences a tempor-

ary reduction in S response rate. While the 2.5-sec.

2

delay during 82 was inconsistent in its effect on the S2

. .- '-.:_.x a ”a. ,, 5

rate of response, such was not the case for its effect on

the S1 rate of response. The delaying of reinforcement

produced an increase in the S re-

for 2.5-sec. during S 1

2

Sponse rate for all subjects and, hence, the group mean.
As with the 2.5-sec. delay, Figure 2 shows that

the 5—sec. delay had not consistent effect on the S re—

2
Sponse rates. The $2 reSponse rates either showed no
change (bird (2655), a permanent increase (bird 622), a
temporary decrease (birds 435 and 389), or a permanent
decrease (birds 32, 2955, and 2549). In terms of S1 re—
sponding, five subjects (birds 32, 435, 2655, 389, and
622) showed an increase in reSponse rate. While two of
the subjects (birds 2955 and 2549) showed no apparent in—
crease in S1 response rate, it should be noted that neither

of these subjects reSponded at a stable rate prior to the

introduction of the reinforcement delay, i.e., to the left

11

of the dashed vertical line. The mean curve shows a tem—

porary decrease in the S response rate and a permanent

2

increase in the S1 reSponse rate.

Figure 3 shows that the lO-sec. delay of rein—
forcement produced either a temporary (birds 466, 40, 2032,
1702, and 1794) or a permanent (birds 263 and 990) reduc-

tion in 82 reSponding. Six subjects (birds 466, 2631,

2032, 1702, 990, and 1794) showed a clear increase in S1

reSponse rate; the remaining subject's (bird 40) S1 re-

Sponse rate curve is only suggestive of an increase in the
last 3 sessions (sessions 38-40). The mean S1 and 82 re-
sponse rate curves for the lO-sec. delay group are similar
to those of the 2.5 and 5—sec. delay groups as the former

shows a permanent increase and the latter a temporary de-
crease.

From Figures 4 and 5 it is apparent that delaying
reinforcement for 120-sec. is similar to nonreinforcement

in that both procedures reduce S reSponding to a near

2

zero level. In terms of S1 reSponse rates, six subjects
(birds 1314, 238, 35, 2518, 889, and 2748) in the 120—sec.

delay group showed an increase, and all subjects in the

extinction group showed an increase. Mean curves show a

12

reduction in 52 rate to near zero and an increase in S1

rate.
In order to determine if the 5 groups differed in

the amount of increase in S1 response rates, the final 10

sessions were collapsed into six 3-day blocks (Table 1).

Thus, the first block is the mean Sl response rate during

the last 3 sessions of the multiple VI VI without rein—
forcement delay. The remaining five blocks represent mean

81 rates during the sessions where reinforcement during

S2 was either delayed or discontinued. The results of a

two—way (trials by groups) analysis of variance with re—
peated measures on one of the factors (trials) are pre—
sented in Table 2. From this table it is quite clear that

the only significant effect was the increase in S response

1

rate over sessions due to the changes in the $2 contin-

gencies. The groups did not differ significantly in the
magnitude of the S1 response rate increase.

Several Pearson product-moment correlations were
computed to determine if the magnitude of the increase in

S1 responding could be predicted by the magnitude of the

decrease in $2 responding. The magnitude of the S1 re—

Sponse rate increase was taken as: (the mean S response

1
rate calculated from sessions 23-25)—-(the highest daily

13

S1 reSponse rate during sessions 26-40). The magnitude

of the 82 response rate decrease was taken as: (the mean

82 response rate calculated from sessions 23-25)--(the

lowest daily S response rate during sessions 26—40). The

2

obtained correlations for each group, as well as the over-

all correlation, are presented in Table 3. These corre-

rflur-m:r:r-T—
. ._.E

lations are small and show no consistency in even their
sign.

Figures 6-8 show the rate of reinforcement during
Sl (denoted by the X's) ands2 @enoted by the closed
circles) for the subjects in the 2.5, 5, and lO—sec. rein—
forcement delay groups. The rate of reinforcement was
computed by dividing the number of reinforcements obtained

during S (or 82) by the total time that the key was illum—

1

inated by S (or $2). In these figures 3 sessions were

1
combined to form each block. The 3 blocks of sessions to
the left of the dashed vertical line were computed from

the last 9 sessions of the multiple VI VI with no rein-
forcement delay (sessions 17—25). The 5 blocks to the
right of the dashed vertical line were computed from the

15 sessions of the multiple VI VI where reinforcemnt during
S was delayed (sessions 26-40). As can be seen in these

2

figures, most subjects showed little change in $2

14

reinforcement rate as a result of either the 2.5, 5, or
lO-sec. delay contingency. Any reductions, such as shown
by birds 4304, 2955, 389, 2631, and 990, were of small
magnitude. The reinforcement rates for the 120—sec. delay
group are not presented since all of these subjects, ex-
cept for bird 1129, showed a reduction in S reinforcement

2

frequency to almost zero.

15

Fig. l.--Response rates to S1 (x's) and $2 (circles)
over sessions. During the 10 sessions to the left of
the dashed vertical line, the schedule of reinforce-
ment was a multiple VI l-min. VI l-min. with no rein—
forcement delay. During the 15 sessions to the right
of the dashed vertical line, subjects continued to be
reinforced on the same multiple VI 1-min. VI l-min.
schedule, but with reinforcement during 82 delayed
for 2.5—sec.

RESPONSES PER MINUTE

l6

2.5-SEC. DELAY

 

 

IIO '-

 

 

 

 

 

I?!

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig. l

J

f_ {@3121

17

Fig. 2.--Response rates to S (x's) and $2 (circles)
over sessions. During the 10 sessions to the left of
the dashed vertical line, the schedule of reinforce-
ment was a multiple VI l-min. VI l-min. with no rein-
forcement delay. During the 15 sessions to the right
of the dashed vertical line, subjects continued to be
reinforced on the same multiple VI l-min. VI l-min.
schedule, but with reinforcement during 82 delayed
for 5-sec.

RESPONSES PER MINUTE

18

5—SEC. DELAY

 

 

 

 

 

'°° WA : . :.
l 1 L 1':

E

 

 

 

 

i?—
< f». .

 

 

 

:WIAWV “J .. ,. 3:;

Fig. 2

19

Fig. 3.--Response rates to S (x's) and S (circles)
over sessions. During the 1 sessions to the left of
the dashed vertical line, the schedule of reinforce-
ment was a multiple VI l-min. VI l—min. with no rein—
forcement delay. During the 15 sessions to the right
of the dashed vertical line, subjects continued to be
reinforced on the same multiple VI l-min. VI l—min.
schedule, but with reinforcement during 82 delayed
for 10—sec.

20

 

lO-SEC. DELAY

 

 

 

«an». «nun». «as». “mm” «a.
a j aaaaa J ‘1 .....
j. .V. 4
(J.
l...
A.
ll ‘\
,YIIIIII'IIIIIIII I'll-I'll.

 

 

 

 

 

 

 

 

we; 322.2 EA. mmmzoammm

/
J]!
A.

 

SESSIONS

3

Fig.

21

Fig. 4.--ReSponse rates to S (x's) and $2 (circles)
over sessions. During the 10 sessions to the left of
the dashed vertical line, the schedule of reinforce-
ment was a multiple VI 1-min. VI 1-min. with no rein—
forcement delay. During the 15 sessions to the right
of the dashed vertical line, subjects continued to be
reinforced on the same multiple VI 1—min. VI l—min.
schedule, but with reinforcement during S2 delayed
for 120-sec.

RESPONSES PER MINUTE

 

IZO-SEC. DELAY

 

an t“
d )-

5? at. M143

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig. 4

 

 

 

23

Fig. 5.—-Response rates to S (x's) and S (circles) over
sessions. During 10 sessions to the left of the dashed
vertical line, the schedule of reinforcement was a mul-
tiple VI l-min. VI l—min. with no reinforcement delay.
During the 15 sessions to the right of the dahsed ver-
tical line, subjects were reinforced on a multiple VI
l-min. extinction schedule.

 

RESPONSES PER MINUTE

24

EXTINCTION

 

 

 

 

 

 

 

 

 

 

 

Finn no
I20 P d '20
I00 P -4 I- /\ d IOO
.0 ’- M'K '1 ‘ I. ‘ \ ‘\{"'I d .0

’M II \x/ x I”
so .- ’H - - / .. to
I"

2° F a d h l; d ‘0
to - q .- co
‘ * kw] .

I m
-I -‘ .0

wk.
\ / « - ”IA‘/H\/\/ .. co
m - " M 4 40
"“ ‘I r- . - co
.. I- -A O
111111111111 V

I- " 1:00
.4 I- A V‘/ - so
P D“ .0

 

 

 

 

 

 

. 5

 

 

 

 

 

4 ISO

'1 M0

'4 Ito

dDO

~00

‘00

:40

‘80

 

 

 

 

25

TABLE 1

RESPONSES PER MINUTE DURING Sl OVER THE SIX 3-DAY BLOCKS (SESSIONS

23-40) FOR THE GROUPS WHICH RECEIVED EITHER NO REINFORCEMENT OR
REINFORCEMENT THAT WAS DELAYED FOR 2.5, 5, 10, OR lZO-SEC. DURING S

 

 

 

2
3-day blocks computed from sessions

Bird
23-25 26-28 29-31 32-34 35-37 38-40
199 44.5 33.6 52.3 48.4 55.4 67.7
3‘ 30 37.9 71.7 74.0 87.6 81.9 59.7
'3 2751 46.4 52.2 59.0 59.8 59.9 72.3
R 609 57.3 64.0 70.6 69.0 73.1 72.7
6 206 90.9 87.1 98.2 95.3 117.3 124.6
3 38 42.3 51.6 59.8 73.5 77.6 83.1
A 4304 57.6 57.7 70.1 84.6 97.7 75.4
a; mean 53.8 59.7 69.1 74.0 80.4 79.4
2655 47.4 48.8 57.7 66.3 76.3 84.7
%. 622 65.0 74.3 92.2 91.9 89.6 97.2
.3 32 40.5 43.1 41.8 43.2 49.7 47.3
6 2955 47.8 47.6 71.5 56.9 31.4 61.9
6 2549 95.0 102.8 114.6 97.2 104.1 99.9
g 389 38.1 42.6 46.3 52.6 57.9 56.7
.5 435 58.0 54.1 89.4 103.6 95.7 87.8
mean 56.0 59.0 73.4 73.1 72.1 76.5
990 43.8 47.6 46.2 56.4 58.4 57.7
5* 466 46.0 60.2 89.4 87.2 77.7 71.6
3 2032 36.1 40.4 40.2 54.2 67.7 '67.7
'9 2631 77.8 95.1 103.9 104.2 119.4 108.9
5 1702 25.6 48.3 66.7 78.0 78.3 74.3
g 40 48.2 48.6 39.8 38.6 48.6 61.9
g, 1794 44.2 44.5 57.6 67.2 80.9 66.2
'9 mean 46.0 55.0 63.4 69.4 75.9 72.6
A, 1314 19.6 28.2 35.4 34.6 35.3 39.4
,3 889 20.5 30.7 35.6 35.9 38.2 38.2
g 2518 32.3 41.3 60.7 71.4 65.0 65.6
. 1129 91.8 83.8 91.4 73.4 82.6 77.9
8 2748 20.5 44.2 65.7 67.6 76.1 67.8
T 35 58.5 92.3 79.1 57.5 66.5 66.2
8 238 81.8 57.4 94.6 87.6 88.0 99.3'
H mean 46.4 54.0 66.1 61.1 64.5 64.9
562 18.2 22.6 43.4 58.2 55.4 71.2
28 67.1 66.1 61.0 81.3 80.5 78.3
8 181 32.4 44.1 . 59.2 61.5 57.8 59.1
13 889 26.9 49.7 72.8 70.9 96.2 83.4
3 1258 35.8 51.8 52.1 47.0 46.8 52.4
‘3 2520 79.0 119.3 96.4 82.2 79.2 87.2
S 4306 80.0 74.9 116.8 150.9 136.3 124.4
mean 48.5 61.2 71.7 78.9 78.9 79.4

 

 

 

 

 

26

TABLE 2

SUMMARY TABLE OF THE ANALYSIS OF VARIANCE ON THE
51 RESPONSE RATES OVER THE SIX 3-DAY BLOCKS
(SESSIONS 23-40) FOR THE GROUPS WHICH RECEIVED
EITHER NO REINFORCEMENT OR REINFORCEMENT THAT
WAS DELAYED FOR 2.5. 5. 10. OR lZO—SEC.

 

 

 

 

 

 

DURING 52.
Source of Variance SS df MS F
Between 85
A (groups) 3299.706 4 824.926 <1
gs within groups 78633.622 30 2621.120
Within Ss
B (blocks of sessions) 17424.125 5 3484.825 27.64l*
AB 1389.556 20 69.477 <1
B x 85 within groups 18911.152 150 126.074
*p < .001
TABLE 3

PEARSON PRODUCT-MOMENT CORRELATIONS BETWEEN THE
MAGNITUDE OF THE INCREASE IN 31 RESPONDING AND
THE MAGNITUDE OF THE DECREASE IN 82 RESPONDING

 

 

2.5-sec. 5-sec. lO-sec. 120-sec. Extinction overall

 

 

 

+.37 +.26 -.02 -.11 -.O3 +.O9

 

27

Fig. 6.--Reinforcement rates during S1 (x's) and $2
(circles) over blocks of sessions (3 sessions per
block). During the 3 blocks of sessions to the left
of the dashed vertical line. the schedule of rein-
forcement was a multiple VI l-min. VI 1-min. with no
reinforcement delay. During the 5 blocks of sessions
to the right of the dashed vertical line. subjects
continued to be reinforced on the same multiple VI
l-min. VI l-min. schedule. but with reinforcement
during S2 delayed for 2.5-sec.

28

2.5-SEC. DELAY

 

 

 

 

 

 

 

 

 

 

 

2345678

BLOCKS OF SESSIONS

345678 I

2

Fig. 6

5. o. 5.
_ — _
.A l
X. l
i I
o X I
X o I
‘.~.‘~‘.
x I
R I v0 I on I X
.. a w m
m X o I v. I 0x I 3 X
14 LE L»
I I v. I X o I X I
X. I X. I In I X0 I
v0 _1 ”I .l v. I. Y. .I
'n I. CA I. .A I M I:
X o X X o X
‘M‘..... .‘.~.~.f~..~.‘ .‘.-.~
! .1 CA I OVA X.
M X I w R I .an 0X I 6 0X I
E - a m
M v. I on X. 0X I
_ _ _ _ _ _ _ _ _ _ _
5. 0. 6 5. 0. 5. 5. 0. 5. 5. 0. 5.
wb323¢ mwa MFZNZNUKOuZEm

'1‘ II In! I I

29

Fig. 7.--Reinforcement rates during 51 (x's) and S2
(circles) over blocks of sessions (3 sessions per
block). During the 3 blocks of sessions to the left
of the dashed vertical line. the schedule of rein-
forcement was a multiple VI 1-min. VI 1—min. with no
reinforcement delay. During the 5 blocks of sessions
to the right of the dashed vertical line. subjects
continued to be reinforced on the same multiple VI
l-min. VI l-min. schedule but with reinforcement
during S2 delayed for 5—sec.

30

5-SEC. DELAY

 

 

J

l

RXXXX

‘ I.

l

 

_
5

LG r—X- x

N
A
m
[T
w

32

 

.1
II

‘NNNNNN‘NNNNNN‘N

I v.

 

.—
50.5

mhzwzwbmouzfim

 

18L
5O

5. 0 5. 5. 0. 5. 5. 0. 5. 5. 0. 5.
_ _ _ _ _ A _ u.
_ Xe. _ I v... .4 x U r
X. I X. I v. I X. L
CA I. VC .J X C I. V. 1
0X 0 X I o X I X
‘M‘..... -~..‘. .‘..‘.~ .‘.‘~.‘
5 9 9 2 .
% X o 8 X M. X o l 2 X
2 I 3 2 6

 

 

2345678

BLOCKS OF SESSIONS

.7

Fig.

31

Fig. 8.--Reinforcement rates during Sl (x's) and S2
(circles) over blocks of sessions (3 sessions per
block). During the 3 blocks of sessions to the left
of the dashed vertical line. the schedule of rein-
forcement was a multiple VI l-min. VI 1-min. with no
reinforcement delay. During the 5 blocks of sessions
to the right of the dashed vertical line. subjects
continued to be reinforced on the same multiple VI
l-min. VI l-min. schedule. but with reinforcement
during 82 delayed for lO-sec.

32

lO-SEC. DELAY

 

 

 

 

 

 

 

 

5 O. 5 5 O. 5 5 O. 5 5. O. 5.
.
X
N I V. I VA C I .A I
CVA VAC I C VA I VC I
X I X J 8 I X L
M I C VA I M I VAC I
2
3 Mu. W 4
O I l l
2 v0. W V. 9 0X m I I
R . I. A L
M I CVA I C VA I C I
V. I CA I v6 I VA C I
‘NNNNNNNNNNNNNNNNNNNNNN NNNNNNN
CA I VC I v. C X
N v. I 6 i I 0 X o I I X. I
a 6 4 3
4. 6
M 00 I c X I X o I 2 l I
_ _ P F _ _ F _ _ b L! _ _ _
5. o. 5. 5. o. 5 5. o. 5. 5 o. 5
w k 3 z _ 2

mun 3523461sz

2345678

BLOCKS OF SESSIONS

345678

 

Fig. 8

DISCUSSION

Behavioral contrast is said to occur if. following

a change in the S reinforcement contingencies the increased

2

S1 response rate changes in a direction away from that main-

tained by S

 

2. The present experiment was conducted to de-
termine whether the delaying of reinforcement during one
component of a multiple VI 1-min. VI 1-min. schedule would
produce behavioral contrast. The relationship between the
duration of reinforcement delay and the amount of behav—
ioral contrast produced. if any. was also examined.

From Figures 1-4 it is quite clear that the delay-
ing of reinforcement during one of the multiple schedule's
components did produce behavioral contrast. 0f the 28 sub-

jects that received delayed reinforcement during S 24

2.
clearly showed behavioral contrast. Three of the four sub—
jects that failed to show behavioral contrast (Birds 1129.
2955. and 2549) also failed to show stable response rates
prior to the introduction of the reinforcement delay. The

responding of the remaining subject (Bird 40) was

33

34

suggestive of behavioral contrast but only during the last
3 sessions of the reinforcement delay contingency.

Using a modified discrete—trial paradigm. where
both S1 and 82 were associated with different durations of
reinforcement delay and where the delay was not associated
with a stimulus change. Keller (1970) also demonstrated be-
havioral contrast. The conclusion that delayed reinforce-
ment produces behavioral contrast. thus. seems to hold
across large variations in procedure.

Groups that received reinforcement delays of 2.5.
5. 10. or 120—sec. or even extinction during S2 did not
show a statistically significant difference in the magni-
tude of behavioral contrast. This failure to find magni-
tude differences may. however. be due more to the between-
subjects design employed than to the absence of a rela-
tionship. Most studies concerned with the amount of be—
havioral contrast (e.g.. Reynolds. 1961b; 1963; Thompson.
1965; Bloomfield. 1967) have employed a within-subjects
design. There were. however. two related reasons for the
present study's between-subjects design. Since behavioral
contrast follows a temporal course and even disappears

after extended training (Terrace. 1966b). a within-subjects

design would require counterbalancing the presentation

35

order of the different delay durations. But. the effect of
a particular duration of reinforcement delay has also been
shown to depend on the manner in which it is introduced
(Ferster. 1953). Thus. a particular reinforcement delay
might have different effects depending on whether it was
presented at full duration (not preceded by shorter rein-
forcement delays) or gradually increased over sessions
(preceded by shorter reinforcement delays).

So.“while there were distinct advantages to using

a betwee-subjects design. such a design may not have been

the most sensitive to differences in the magnitude of be-

 

havioral contrast. It should be noted. however. that a
recent between-subjects experiment (Taus & Hearst. 1970)
was able to show statistically significant differences in
the amount of behavioral contrast. ‘

A second possible reason for the failure to find a
relationship between duration of delay and the amount of
behavioral contrast concerns the value of the baseline VI
schedule. Reynolds (1963) suggests that the reinforcement
rate during the unaltered component in part determines the
magnitude of behavioral contrast. Perhaps. the use of a
VI 3-min. baseline. which Reynolds has shown to be quite

sensitive to behavioral contrast. would have been more

36

appropriate. A third possibility is that the function
relating delay duration to the amount of behavioral con-
trast reaches asymptote at or before the minimum delay
duration of the present study (2.5—sec.). Indeed. a more
positive conclusion might have been possible if some
shorter delays had been included.

Results of the present study also have important
implications for conditions necessary to the production of
behavioral contrast. From Figures 6-8 it is clear that
behavioral contrast can be produced without reducing the
reinforcement rate during the altered component. This
finding is in agreement with several studies that were
specifically designed to hold reinforcement rate constant
during the altered component (e.g.. Terrace. 1968; Weisman.
1969; 1970).

Terrace (1971a) maintains that behavioral contrast
is a by-product of some emotional responses that are pro-
duced by the subject's reduced and inhibited responding
during the altered component. 0f the 21 subjects in the
2.5. 5. and lO-sec. delay groups. four showed no reduction
in S2 response rate; however. all of these showed behav-

ioral contrast. Nine subjects showed only a temporary re—

duction in S2 responding and eight of these showed

37

behavioral contrast. 0f the eight that did permanently
reduce $2 responding. six showed behavioral contrast.
Clearly. a permanent reduction in responding during the
altered components is not necessary to the production or
maintenance of behavioral contrast. The present data
further suggest that behavioral contrast may occur regard-
less of any_reduction in responding during the altered
component.

While only 4 subjects did show behavioral contrast
without a reduction in responding during the altered com-
ponent. it does seem that Terrace's position needs some
modification. Keeping within the rubric of his emotion—
ality theory. it may be that the response rate reduction.
like behavioral contrast. was just a frequent by-product
of some emotional responses elicited by the reinforcement
delay. In support of this interpretation. it should be
noted that in many cases the response rate reduction. like
behavioral contrast (Terrace. 1966b). disappeared with
further training. Extended training (Terrace's subjects
received 60 sessions) might have shown the response rate
reduction to be temporary in all subjects.

If one accepts the emotionality interpretation of

behavioral contrast. there still remains the problem of

38

specifying the necessary and sufficient conditions that
produce the frustration or other emotional responses.
Guttman's suggestion that the weaker of two reinforcement
schedules can become "functionally negative." while in-
trinsically appealing. is not very useful. There are no
specified or readily apparent criteria for classifying one
of two schedules of equal reinforcement frequency as the
weaker. For example it is unclear why a DRO should be con-
sidered weaker than a VI of equal reinforcement frequency.
and. yet. it is quite clear that changing a multiple VI VI
to a multiple VI DRO does produce behavioral contrast
(Weisman. 1970). Premack's (1969) suggestion that

Contrast results if and only if there is a change

in the aversiveness associated with one of the

components in the schedule. [p. 136]
may be the most workable hypothesis at present. This hy-
pothesis is. of course. quite similar to Bloomfield's
(1969) contention that behavioral contrast is produced by
"a worsening of conditions" during one of the components.
If behavioral contrast occurs the stimulus associated with
the altered component should possess all the properties
of other aversive stimuli--subjects should learn a new

response to escape from the stimulus and the stimulus

should serve as an elicitor of aggression and as a

39

punisher of ongoing behavior. Likewise. if no contrast is
observed. these properties should not be present. Although
a beginning has been made in studying multiple schedules
within these different paradigms (e.g.. Rilling. Askew.
Ahlskog. & Kramer. 1969; Rilling. Kramer. & Richards. 1971;
Terrace. 1971b; Weisman & Premack. 1966). more research
will be required before a definitive conclusion can be

made.

LIST OF REFERENCES

LIS T OF REFERENCES

Amsel. A. The role of frustrative non-reward in noncon-
tinuous reward situations. Psychological Bulletin.
l958._§5. 102-119.

Bloomfield. T. M. Behavioral contrast and relative rein—
forcement frequency in two multiple schedules.
Journal of the Experimental Analysis of Behavior.
1967..lg. 151-158.

 

Bloomfield. T. M. Behavioral contrast and the peak shift.
In R. M. Gilbert and N. S. Sutherland (Eds.)
Animal discrimination learning. London: Academic
Press. Pp. 215-241.

 

Brethower. D. M. and Reynolds. G. S. A facilitative effect
of punishment on unpunished behavior. Journal of
the Experimental Analysis of Behavior. 1962. 5.
191-199.

 

Brown. J. S. and Farber. I. E. Emotions conceptualized as
intervening variables-~with suggestions toward a
theory of frustration. Psychological Bulletin.
1951. 48, 465-495.

Dunham. P. J. Contrasted conditions of reinforcement: a
selective critique. Psychological Bulletin. 1968.
62, 295-315.

Ferster. C. B. Sustained behavior under delayed reinforce-
ment. Journal of Experimental Psychology. 1953.
45, 218-224.

Guttman. N. Generalization gradients around stimuli asso-
ciated with different reinforcement schedules.
Journal of Experimental Psychology. 1959. 58, 335-
340.

40

41

Holder. W.. Marx. M.. Holder. Elaine. and Collier. G.
Response strength as a function of delay of reward
in a runway. Journal of Experimental Psychology.
1957. 2;. 316-323.

Keller. J. V. Behavioral contrast under multiple delays of
reinforcement. Psychonomic Science. 1970. 39_(5).
257-258.

Premack. D. On some boundary conditions of contrast. In
J. Tapp (Ed.). Reinforcement and behavior. New
York: Academic Press. 1969. Pp. 120-145.

Renner. K. E. Delay of reinforcement: A historical re—
view. Psychological Bulletin. 1964. §;, 341-361.

Reynolds. G. S. Behavioral contrast. Journal of the Ex-
perimental Analysis of Behavior. 1961a. 4, 57-71.

 

Reynolds. G. S. Relativity of response rate and reinforce-
ment frequency in a multiple schedule. Journal of
the Experimental Analysis of Behavior. 1961b. 4,
179-184.

 

Reynolds. G. S. Contrast. generalization. and the process
of discrimination. Journal of the Experimental
Analysis of Behavior. 1961c. 4, 289-294.

Reynolds. G. S. Some limitations on behavioral contrast
and induction during successive discrimination.
Journal of the Experimental Analysis of Behavior.
1963. g. 131-139.

Reynolds. G. S. and Limpo. A. J. On some causes of behav—
ioral contrast. Journal of the Experimental Anal-
ysis of Behavior. 1968. 44, 543-547.

Rilling. M.. Askew. H. R.. Ahlskog. J. E.. and Kramer.
T. J. Aversive properties of the negative stimulus
in a successive discrimination. Journal of the
Experimental Analysis of Behavior. 1969. 44, 917-
932.

42

Rilling. M.. Kramer. T. J.. and Richards. R. W. Aversive
properties of the negative stimulus during learning
with and without errors. Paper presented at the
Eastern Psychological Association meeting. New
York. 1971.

Staddon. J. E. R. Multiple fixed-interval schedules:
Transient contrast and temporal inhibition. Jour—
nal of the Experimental Analysis of Behavior. 1969.
4;, 583-590.

Taus. Sharron E. and Hearst. E. Effects of intertrial
(blackout) duration on response rate to a positive
stimulus. Psychonomic Science. 1970. 12_(5).
265-266.

 

Terrace. H. S. Stimulus control. In W. K. Honig (Ed.).
Operant behavior: Areas of research and applica-
tion. New York: Appleton-Century-Crofts. 1966a.
Pp. 271-344.

Terrace. H. S. Behavioral contrast and the peak shift.
Journal of the Experimental Analysis of Behavior.
1966b..2. 613-617.

Terrace. H. S. Discrimination learning. the peak-shift.
and behavioral contrast. Journal of the Experi-
mental Analysis of Behavior. 1968. 44, 727-741.

 

Terrace. H. S. By-products of discrimination learning.
In Bower and Spence (Eds.). Learning and motiva-
tion. 1971a. y, in press.

 

Terrace. H. S. Escape form S-. Learning and motivation.
1971b. in press.

 

Thompson. D. M. Punishment by SD associated with fixed
ratio reinforcement. Journal of the Experimental
Analysis of Behavior. 1965. g, 189—194.

Weisman. R. G. Some determinants of inhibitory stimulus
control. Journal of the Experimental Analysis of
Behavior. 1969. 43, 443-450.

 

 

43

Weisman. R. G. Factors influencing inhibitory stimulus
control: differential reinforcement of other be-
havior during discrimination training. Journal
of the Experimental Analysis of Behavior. 1970.
44, 87—91.

Weisman. R. G. and Premack. D. Positive and negative
generalization gradients depending on the relative
rates of reinforcement in two components of a
schedule. Paper read at Psychonomic Science meet-
ing. St. Louis. 1966.

Wilkie. D. M. On some determinants of behavioral contrast.
Unpublished doctoral dissertation. University of
Manitoba. 1970.

I‘IIC

AIIIIIIIIIIIAIIIIII“