A HA8?!" STRENGTH ANALYSES OF THE
RELATDNSH}? EETWEEN RESPONSE EFFSR? AND
REESTANCE "2'0 EXTENC‘HON

 

fixed: fiat £1; Mama :5 333g 91
MlCHiGAN 9351?? UNEV’ERSiTY
‘i’homes £02m Sfachnik
W53

TH ESlS

This is to certify that the

thesis entitled

A HABIT STRENGTH ANALYSIS OF THE
RELATIONSHIP BETWEEN RESPONSE EFFORT
AND RESISTANCE TO EXTINCTION

presented by

Thomas J. Stachnik

has been accepted towards fulfillment
of the requirements for

M. degree in mow

' Major essor 71/? ‘

Date é _Q/~ é 3

 

0-169

 

LIBRARY

Michigan State
University

 

A HABIT STRENGTH ANALYSIS OF THE
RELATIONSHIP BETWEEN RESPONSE EFFORT
AND RESISTANCE TO EXTINCTION

By

Thomas John Stachnik

AN ABSTRACT OF A THESIS

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

DOCTOR OF PHILOSOPH

Department of Psychology

1963

ABSTRACT

A HABIT STRENGTH ANALYSIS OF THE
RELATIONSHIP BETWEEN RESPONSE EFFORT
AND RESISTANCE TO EXTINCTION

by Thomas John Stachnik

This study was designed to determine the importance of
effort as a variable in experimental extinction. Current
textbooks either admit that the importance has not been
determined, or assume it is important on the basis of data
which cannot be considered definitive. Contrary to previous
studies, the present work includes the assumption that a
heavy-bar habit subsumes a light-bar habit, but that the
converse is not true. Seven groups of animals (rats) were
given various amounts of training on differentially weighted
bars in a Skinner box and then extinguished to a 5 minute
no-response criterion with the following results:

1. Contrary to the prediction of Ir, the difference
in resistance to extinction between groups trained and
extinguished on a light bar and those trained and extin-
guished on a heavy bar is non-significant when an asymptote
has been reached and operant level is taken into account.

2. A partial replication of a previous study in which
effort and extinction were shown to be directly related

indicated that an asymptote had not been established, and

Thomas John Stachnik
that the findings are better interpreted in terms of differ-
ential habit strengths.

3. Also contrary to the prediction of Ir, animals
trained on a heavy bar and extinguished on a light bar show
a facilitation When compared with animals trained and extin-
guished on a light bar.

4. The facilitation resulting from heavy bar training
and light bar extinction was significantly reduced by insert-
ing light bar responses prior to extinction. This suggests
that the facilitation is a function of a contrast effect
and not the magnitude of the acquisition task. The incidence
of contrast effects in various eXperimental situations was
briefly reviewed and the lack of information of the para-
meters involved was pointed out.

5. An interference theory of extinction was suggested
as being more parsimonious and appropriate in accounting

for the present findings.

A HABIT STRENGTH ANALYSIS OF THE
RELATIONSHIP BETWEEN RESPONSE EFFORT
AND RESISTANCE TO EXTINCTION

By

Thomas John Stachnik

A THESIS

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

DOCTOR OF PHILOSOPHY

Department of Psychology

1963

(7 16}! <1"?
7/67”

ACKNOWLEDGMENTS

The author gratefully acknowledges the assistance
given him by his major professor and friend, M. Ray Denny,
without whose ideas and direction this study would not have
been possible. I am also grateful to the other members of
my guidance committee for their help and encouragement:
Doctors Donald Johnson, James Karslake, William Kell,
and Stanley Ratner. David Silkiner, who generously gave
his time to check the statistical work, also deserves my
thanks. Lastly, I want to formally express my gratitude
to my wife, Kay, who not only typed all drafts of the
present paper, but graciously accepted many husband-less

evenings while the data were being collected.

11

TABLE OF CONTENTS

Chapter Page
I. INTRODUCTION . . . . . . . . . . . . . . . . . . . 1
II. METHOD . . . . . . . . . . . . . . . . . . . . . 13
III. RESULTS . . . . . . . . . . . . . . . . . . . . 22
IV. DISCUSSION . . . . . . . . . . . . . . . . . . . 30
V. SUMMARY . . . . . . . . . . . . . . . . . . . . 42

REFERENCES 0 O O O O O O O O O O O O O O O O O O 44

111

LIST OF TABLES

Page
Summary of groups and experimental design . . . . 13

A comparison of the mean number of responses
made during extinction by Groups H H 120
and H H 20 O O O I O O O O O O O O O O O O O O O 23

A comparison of the mean number of responses
made during extinction by Groups L L 200,
and H H 120 and H H 200 combined . . . . . . . 23

A comparison of the mean number of responses
made during extinction by Groups L L 200
and H L 1 2O 0 O O O O O O O O O O O O O O O O O 25

A comparison of the mean number of responses
made during extinction by Groups H L 120
and P111 L 1 2O 0 O O O O I I O O O O O O O O O O 25

A comparison of the mean number of responses
made during extinction by Groups L L 200
and PE L 1 20 O 0 O O O O O O O O O O O O O O O 27

A comparison of the mean number of responses

made during extinction by Groups LMH L 90

and LMH H 90 . . . . . . . . . . . . . . . . . 27
A comparison of the mean number of responses

made during extinction by Groups H H 120

and LNH H 90 o o o o o o o o o o o o o o o o o 28

Summary of comparisons made ... . . . . . . . . . 29

iv

I. INTRODUCTION

It is painfully clear to anyone who has ever chopped
wood, done calisthenics, or run long distances that some
behaviors fatigue muscle tissue to the point where continued
responding becomes aversive. Any man on the street knows
this to be true. However, the effect of responding 22; £2,
1.6., when lesser amounts of work are involved and tissue
is not fatigued, is not nearly so well understood. In
fact, although this problem has received considerable
attention from experimental psychologists, it remains
unresolved.

It is of interest to note how the relationship between
the effort variable and resistance to extinction has been
treated in recent textbooks. Twenty years ago Hull (1943)
introduced the constructs of reactive inhibition (Ir) and
conditioned inhibition (sIr) in an attempt to account for a
variety of learning and inhibitory phenomena. These con-
structs emphasized the presumed inhibiting value of response,
22;,gg, and generated extensive research. Many of the find-
ings from this research failed to confirm the predictions of
Ir and sIr, yet current textbook treatments of the effort
variable place a great deal of importance on effort as a

variable in experimental extinction.

2

A case in point is a recent book by Donald Lewis (1963).
His discussion of the effort variable is largely a summary
of the findings of Capehart, Viney, and Hulicka (1958) and
their conclusion that "the number of extinction responses
is an almost linear function of the effort required to push
the bar" (1958, p. 209). Another recent text by Howard
Kendler (1963) refers to a 1943 study by Mowrer and Jones,
accompanied by a graph which also shows resistance to ex-
tinction as an inverse linear function of the effortfulness
of the task. However, both of these studies, which will
later be discussed at length, are open to alternative inter-
pretations and cannot be considered definitive.

Probably the most comprehensive text dealing with learn-
ing and inhibitory phenomena is Gregory Kimble's revision
of Hilgard and Marquis' Conditioning agd_Learning (1961).
In the discussion of extinction and the effortfulness of
the response, Kimble lists a number of studies which have
demonstrated that increased effort resulted in faster extinc-
tion, and also mentions that one study yielded negative
results. He follows with this statement:

Although the evidence seems clear on the point

that effort does influence extinction, the form of

the function, and the importance of this variable,

are still in doubt. This is because the problems

of manipulating effort, but no other variable, in

studies of extinction are greater than might have

been anticipated (1961, p. 285).
One of the problems he subsequently refers to is that in

training an animal to press a heavy bar it is first

3

necessary to teach it to press a lighter bar, and that this
may result in animals extinguished on a light bar receiving
more training than those extinguished on a heavy bar.

Current textbooks, then, either admit that the importance
of the effort variable has not been determined, or assume it
is important on the basis of data which cannot be considered
definitive. The present study will examine the effect
of the effort variable, hopefully with the proper controls,
in an attempt to determine the importance of effort as a
variable in experimental extinction. Specifically, the role
of effort is being evaluated in a free-responding, operant

situation.

A Brief History

The concept of Ir which generated so much effort-
variable research was defined by Hull as, ". . .a condition
or state which acts as a primary negative motivation in that
it has an innate capacity to produce cessation of the
activity which produced the state" (1943, p. 278). Two
explicit characteristics of Ir were also formulated, one of
which is the relationship between Ir and the work or effort
involved in responding. "The net amount of functioning
inhibitory potential resulting from a sequence of reaction
evocations is a positively accelerated function of the amount
of work (W) involved in the performance of the response in

question" (1943, p. 278). The second characteristic of Ir

explicitly stated is that it ". . .diminishes progressively

4

with the passage of time according to a simple decay or
negative growth function" (1943, p. 281). This suggests
that Ir is like fatigue in that it is reduced through rest.

Many of the studies designed to test the predictions
of Ir have occurred in the framework of the spontaneous
alternation phenomenon. Alternation is said to have
occurred when on the second trial in a T-maze the animal
enters the arm opposite the one visited on trial one; either
or both arms of the T-maze may or may not contain a reward.
When one reviews this work, two things are apparent: first,
it is extensive, and second, the results are conflicting,
often with two findings directly contradicting each other.
An example of the latter is the 1948 study by Solomon
(1948a) and a study by Walker, Dember, Earl, Fawl, and
Karoly (1955). Solomon inclined the goal arms of a T-maze
160 from the horizontal, which meant more work was involved
in the response, more Ir presumably generated, and hence
more alternation-~and in this instance the prediction of
Ir was supported. Walker gt 3;. inclined the goal arms to
45°, which should have resulted in even more alternation,
but such was not the case; in fact, the frequency of alter-
nation was lowered.

The most direct refutation of an Ir prediction in a
spontaneous alternation framework occurred in similar
studies by Montgomery (1952) and Glanzer (1953). Both

experimenters utilized a cross maze:

 

 

 

Rats were placed in the starting alley at point A and
allowed to turn into either one of goal arms C or D (B was
blocked off). Following this, the animals were immediately
taken out and placed in the starting alley at point B and
again allowed to turn into either goal arm C or D (A was
blocked off). If on trial one, the animal turned right
into goal arm D, Ir would predict that on trial two the
animal would turn left, returning into goal arm D, 1.9.,
would alternate responses rather than goal arm visited.

In both studies the Ir prediction was wrong--the animals
repeated the previous response and visited the yet unex-
plored goal arm.

Varying intertrial interval has also served as a test
of Ir, since the theory predicts that the inhibition dimin-
ishes progressively with the passage of time. The most
extensive attempt to test this was conducted by Walker
(1956), again using the amount of alternation in a T-maze
as the dependent variable. Reactive inhibition predicts

that the probability of alternation should decrease as the

6
intertrial interval is increased, i.e., with a long inter-
trial interval, the Ir presumably generated by a previous
response should dissipate so the repetition of a previous
response is possible. Walker used 21 values of intertrial
interval ranging from 1 to 300 minutes, and his results
generally supported the theory. But there was no clear
relationship between amount of alternation and length of
intertrial interval. The amount of alternation varied in
a random fashion around a value of 75 percent for intervals
up to about 60 minutes; between 60 and 90 minutes there
was a sharp drop to chance level. At longer intervals
there were some below-chance levels of alternation.

Maatsch (1955) gave rats extended training on both
spaced and massed trials in a straight alley maze and com-
pared the predictions made by Ir with those of an interference
theory. He found that none of the major predictions drawn
from Ir were confirmed. If given two trials in a straight
alley maze with the second trial immediately following the
first, the Ir generated on trial one should be present at
the beginning of the second trial and should inhibit per-
formance. The data indicated that no inhibition was present
as a result of the first trial.

The effort variable has also been manipulated in more
complex instrumental situations. Aiken (1957) had rats push
a weighted swinging door to obtain food. Half of the S3

were trained under low effort conditions and half under high.

 

A if {il‘f

7

During extinction each group was divided in two and extin-
guished on either low or high effort. The results indicated
that on the first extinction day, the animals trained and
extinguished on high effort actually made more responses
than those trained and extinguished under low effort, though
not significantly so. However, the results are equivocal
since on the second day of extinction the reverse was true;
in this case, the difference was significant.

The Skinner box has also provided a convenient means
of testing the predictions of Ir, since the bar loading,
and hence the effort involved in responding, can easily be
manipulated by the experimenter. But the relationship
between effort and resistance to extinction remains unclear
because most of these studies failed to include the proper
controls, particularly for the habit strength of the
response in question.

In the 1943 Mowrer and Jones study previously referred
to, 30 rats were trained with variously weighted levers,
so that during acquisition every rat made 180 responses on
a 5 gm. weight, 20 responses each on 30 and 55 gm., and 60
responses on an 80 gm. weight. Ten animals were then ran-
domly assigned to each of three groups and extinguished on
either 5 gm., 42.5 gm., or 80 gm. bar loadings. The
results indicated that resistance to extinction was a de-
creasing monotonic function of the bar loading. Hull called

their report "convincing evidence" (1943, p. 279). and

8
Solomon regarded it to be "unequivocal in supporting some
kind of formulation which takes into account the negative
motivating aspect of work or effort" (1948a, p. 93). How-
ever, since the rats received more reinforcements on the
light bar than they did on the heavy during acquisition,
differences in habit strength would be a more parsimonious
explanation of the extinction differences than effort and
Ir.

Two other studies, both purporting to have controlled
for habit strength, report that in a lever-pressing situa-
tion, resistance to extinction is an inverse linear function
of the effortfulness of the task; yet neither of these
studies can be considered definitive. Applezweig (1951)
began his experiment with 179 rats, but eventually discarded
79 of them, mostly those trained on the heavy bar. This
was true because animals have a weak initial tendency to
press a heavy bar, and this resulted in a non-random
selection of Se in some groups. His results are also ques-
tionable because the design of the study assumed that if a
rat is trained to depress a light bar, it depresses heavier
bars with the same facility that rats trained to depress
heavy bars depress lighter bars. One has only to observe
the behavior of the animals to conclude that this assumption
is not tenable. A rat trained to depress a heavy bar will
easily depress all lighter bars, but the reverse is not true.

Unless the loadings on the bar are graduated, the rat

9
trained on a light bar and then switched to a heavy bar will

extinguish quickly if the extinction criterion is, say, five
minutes of no responding. Maatsch, Adelman, and Denny
(1954) make this point in their study which showed no rela-
tionship between effort and extinction. In the Applezweig
study, the animals switched to heavy bars during extinction
did extinguish quickly; but there was little or no relation-
ship between the bar loading and resistance to extinction
when the animals were extinguished under the same condition
they were trained under. When the data were combined,
however, it appeared that resistance to extinction was a
function of the effort of the task.

In the second study, Capehart, Viney, and Hulicka (1958)
equated the number of reinforcements at each bar loading,
but it can be argued that they failed to control for habit
strength because of the same assumption, namely, that rats
trained to push light bars push heavier bars with equal
facility. After three days of pre-training, which consisted
of habituation to the apparatus and the association of food
with the "click" of the food—delivery mechanism, each ani-
mal made 15 reinforced responses per day for six consecutive
days on bar loadings of 5, 40, 70, 5, 70, and 40 gm. Thus,
each animal made a total of 90 reinforced responses. But
it is quite legitimate to consider the reinforcement of a
bar press of a given loading weight to reinforce the bar-

pressing habit for all bars of equal or lesser weight, but

1O
ngt_for bars of greater weight. This would mean that each
animal had 90 reinforcements of the habit for the 5 gm. bar
(30 on 5, 30 on 40, 30 on 70) but only 60 reinforcements
of the habit for the 40 gm. bar (30 on 40, 30 on 70) and
only 30 reinforcements of the habit for the 70 gm. bar (30
on 70 only). In this specific sense, all studies which
have presumably demonstrated the inverse relationship
between effort and resistanoeto extinction can be considered
confounded with a differential habit strength variable.

In summary, experiments designed to test the predictions
from Ir have been somewhat inconclusive, although largely
nonsupportive. Studies employing the Skinner box have been
contradictory and particularly inconclusive because of a
lack of control of a number of variables. The most con-
spicuous is the habit strength of the response in question.
The present study is a special attempt to determine the
relationship between effort and extinction with habit strength

controlled.

Predictions To Be_$ested

The predictions which follow are based on a habit
strength interpretation which includes the assumption that
a heavy-bar habit subsumes a light-bar habit, but that the
converse is not true. These predictions can be better under-
stood once the procedures have been described, and therefore
will be reiterated in the description of the groups. Some

of these predictions, e.g., number 6, were not formulated

11
in the original design, but evolved as the experiment pro-
gressed.

1. There will be no significant difference in the
number of responses to extinction between animals trained
and extinguished on a light bar (Group L L 200) and those
trained and extinguished on a heavy bar (Group H H 120).
Reason: Habits are equal.

2. There will be no significant difference in the
number of responses to extinction between animals given 120
heavy-bar reinforcements (Group H H 120) and those given
200 reinforcements (Group H H 200) when both are extin-
guished on a heavy bar. Reason: Both groups at asymptote.

3. Animals given 90 reinforcements on varied-bar
loadings (Group LMH L 90) and then extinguished on a light
bar will make more responses during extinction than those
similarly trained, but extinguished on a heavy bar (Group
LMH H 90). Reason: Unequal habit strengths.

4. Animals given 120 reinforcements on a heavy bar
(Group H H 120) and then extinguished on a heavy bar will
make more responses during extinction than those given 90
reinforcements on varied bar loadings and extinguished on
a heavy bar (Group LMH H 90). Reason: Unequal habit
strengths.

5. Animals trained on a heavy bar and extinguished on
a light bar (Group H L 120) will make more responses during
extinction than those trained on a light bar and extinguished
on a light bar (Group L L 200). Reason: Previously found

12
by other investigators. Interpretation: Differential habit
strength or behavior contrast.

6. a) If the facilitation in Group H L 120 is a func-
tion of greater habit strength resulting from increased
response effort, there will be no significant differencein
the number of extinction responses when animals are trained
similarly but given light-bar reinforcements just prior
to extinction (Group HL L 120). Reason: The interspersed
light-bar reinforcements will not affect habit strength.

b) If the facilitation in Group H L 120 is a
function of a contrast in acquisition and extinction stimu-
lation, animals trained similarly but given light-bar rein-
forcements just prior to extinction (Group HL L 120) will
make less responses during extinction. Reason: The inter-
spersed light-bar reinforcements will minimize the contrast

effect.

II.

METHOD

Table 1 below summarizes the groups used and the com-

parisons made among them.

for each group indicate the following.

The letter and number designations

The first letter

or group of letters refers to light (L), medium (M), or

heavy (H) bar training; the second letter indicates the bar-

loading during extinction; and the number refers to acquis-

ition reinforcements.

Table 1.--Summary of groups and experimental design.

 

Group N No. and Bar-Loading Bar-Loading Comparisons
of Reinforcements During
Extinction
L L 200 14 200 on 5 gm. 5 gm. H H 200 vs.
H H 120-H H 200
H H 120 10 80 on graduated load- 80 gm. H H 120 vs.
ings, 120 on 80 gm. H H 200
H H 200 10 80 on graduated load- 80 gm.
ings, 200 on 80 gm.
H L 120 10 80 on graduated load- 5 gm. H L 120 vs.
ings, 120 on 80 gm. H H 200
HL L 120 10 50 on graduated load- 5 gm. HL L 120 vs.
ings, 120 on 80 gm., H L 120;
30 on 5 gm. HL L 120 vs.
H H 200
LMH L 90 1O 30 on 5 gm., 30 on 40 5 gm. LMH L 90 vs.
gm., 30 on 70 gm. LMH H 90
LMH H 90 1O 30 on 5 gm., 30 on 40 70 gm. LMH H 90 vs.

gm., 30 on 70 gm.

H

H 120

 

13

14

Subjects

Seventy-four male rats, both albino and grey hooded,
served as gs. The animals were approximately 90-140 days
old at the beginning of the experiment and were from the
colony maintained by the Psychology Department at Michigan
State University. Possible age and strain differences were
controlled for by counterbalancing the Se across all groups
between which comparisons would be made. The data indi-
cated, however, that neither of these variables had any

observable effect upon performance in the present study.

Apparatus

The apparatus employed was a Skinner box 14 in. long,
14 in. deep, and 9 1/4 in. wide. The sides were a clear
plexiglass, while the bar, which was 4 in. long, 3/4 in.
wide and 1/8 in. thick, was of red plexiglass and protruded
1 in. into the box at a height of 3 1/2 in. from the floor.
A metal food cup was located just to the right of the bar,
1 1/8 in. from the floor. The bar could be counterweighted
so that the effort required to activate the food-delivery
mechanism could be set as low as 5 gm., as high as 80 gm.,
or at any value in between. An electric counter recorded
all bar-press responses which activated the feeding mechan-
ism. The food reward used was a 45 mg. pellet, 4 mm. by 3.3

mm., made by the P. J. Noyes Company.

Training
All §s were kept on a reduced diet of 8 gm. of Wayne

15

Lab Blox in their individual home cages for a period of
8 to 10 days prior to beginning the experiment. Water was
available at all times, and the animals were handled for
3 min. periods daily. The §s were under approximately 21 hr.
of food deprivation at the beginning of each experimental
session, and were fed in their home cages at the end of
each session. They received 8 gm. of food in addition to
the pellets obtained during training.

Training was similar for all Se in Groups L L 200,
H H 120, H H 200, H L 120, HL L 120. At the beginning of
training for these Se, the bar required only a 5 gm. de-
pression to activate the delivery mechanism. Three food
pellets were available in the food cup. After these pellets
were eaten, g manually supplied approximately 15 additional
pellets to associate the click of the feeding mechanism with
the arrival of food in the cup. Then, by using the method
ofsuccessive approximation, E shaped the bar-press response
until each § had made approximately 25 reinforced responses,
which terminated the session. The number of reinforced
responses allowed per session varied slightly according to
the bar loadings and total responses designated for each
experimental group (see Table 1), but the number was never
less than 25 or more than 29. The bar loading during sub-
sequent sessions was gradually increased until the §s of
Groups H H 120, H H 200, H L 120, and HL L 120 were able to

depress an 80 gm. bar per the schedule given in Table 1.

16
Each §.was extinguished on the day following the completion
of acquisition to a 5 min. no-response criterion.

The training for Groups LMH L 90 and LMH H 90 was
slightly different, since they represented a partial repli-
cation of the Capehart, Viney, and Hulicka study (1958).
Their procedure for the grommsused was followed exactly.
After 8-10 days of reduced diet and handling, §s received
three days of preliminary training during which time the bar
was removed from the apparatus. On the first day each g was
placed in the apparatus for 20 min. and allowed to eat food
pellets placed in the food cup. On days 2 and 3, E oper-
ated the feeding device so that the audible click of the
device was associated with the arrivalcf food. During the
next six days, §_was required to make 15 responses per day
on bar loadings of 5, 40, 70, 5, 70, and 40 gm. respectively.
The animals were not shaped during this time, nor was the
bar loading graduated, i.e., the day following the first 15
responses on the 5 gm. loading, the loading was increased
to 40 gm. with no intermediate weights. gs therefore made
a total of 90 reinforced responses and upon completing these
were divided into two groups equated in terms of the mean
response latency of the last 45 acquisition responses. One
group was extinguished on a 5 gm. loading and the other on
a 70 gm. loading on the day following the completion of

acquisition trials to a 5 min. no-response criterion.

17

Description of Groups and Epperimental Comparisons

The comparisons referred to in the following para-
graphs are between the mean number of responses made by
each group during extinition to the same extinction criterion.

Group L L 290.--All gs were required to make a total
of 200 reinforced responses, all of which were on a light
bar loading (5 gm.), and were then extinguished on a light
bar. It should be noted that 14 animals were used in this
group, whereas in all other groups N = 10. This was the
only group to receive training solely on a 5 gm. bar.

§poup H H 120.--All gs were required to make 80 rein-
forced responses on lighter bar loadings followed by 120
responses on a heavy bar (80 gm.); extinction followed on a
heavy bar. Although Group L L 200 received 200 responses
at the bar loading on which they were extinguished and Group
H H 120 only 120, it was assumed that an asymptote had been
reached by 120 reinforcements. Since the habit strengths
were equivalent, Groups L L 200 and H H 120 should make the
same number of responses during extinction. The prediction
derived from Ir theorizing would be that Group L L 200
would be more resistant to extinction than Group H H 120.

figggp H H 200.—-The purpose of this group was to check
on whether an asymptote had actually been reached at 120
acquisition responses in Group H H 120. §s were required
to make 200 responses on the heavy bar preceded by 80 pre-

liminary trials as in Group H H 120. If there were no

18

difference in the number of responses made during extinction
in Groups H H 120 and H H 200, the assumption of equivalent
habit strength with 120 and 200 responses would be tenable.
Furthermore, the data from Groups H H 120 and H H 200 could
be combined for a comparison with Group L L 200. Habit
strength prediction: No significant difference.

Group H L 120.--§s received the same training as Group
H H 120, i.e., 120 responses on the heavy bar preceded by
80 responses on lighter bars, but were then extinguished
on the llgpp bar. Group H L 120 was compared with Group
L L 200 as a further check on the concept of Ir. Whereas
previous studies have indicated that heavy-bar training and
light-bar extinction results in a facilitation in extinction
responding, the Ir prediction would be no significant differ-
ence between the two groups since the effort required
following the termination of reinforcement is identical.

Gpoup HL L 120.--This group was included when a marked
facilitory trend was observed in the first'Mo or three
animals to be run in Group H L 120. It was argued that the
facilitation was due either to a behavior contrast effect
or to greater habit strength in H L 120 resulting from the
greater magnitude of effort involved in the acquisition of
the bar-pressing response. In an attempt to identify the
relevant variable, Se in Group HL L 120 received only 50
preliminary reinforcements (instead of 80 as in H H 120,
H H 200, and H L 120) prior to making 120 responses on the

heavy bar. The 120 reinforcements on the heavy bar were

19
followed by 30 reinforcements on the light bar and then by
extinction on the light bar. The 30 interspersed reinforce-
ments on the light bar was an attempt to eliminate or mini-
mize the contrast effect. Comparisons were then made between
Groups H L 120 and HL L 120 and between Groups L L 200 and
HL L 120. If Group HL L 120 made significantly fewer
responses during extinction than Group H L 120, the facil-
itation would be attributable to a contrast effect. If
no difference were observed, the facilitation might better
be accounted for by an increase in habit strength resulting
from the greater magnitude of the acquisition task (Jaynes,
1950). A comparison between Groups L L 200 and HL L 120
would be in order if the facilitation was a result of a
contrast effect. If contrast effect has been eliminated,
there should be no significant difference between L L 200
and HL L 120.

Groups LMH L 20 and LMH H 20 (Capehart, Viney and
Hulicka).--§s made a total of 90 responses, 30 at each of
three bar loadings: 5 gm., 40 gm., and 70 gm. They were
then divided into two groups, LMH L 90 extinguished on a
5 gm. bar and LMH H 90 on a 70 gm. bar. As previously
argued, the habit strength of the bar-press response has
presumably not been equated in these two groups. If a
reinforcement on a bar of a given loading can be considered
a reinforcement on all bars of equal or lesser weight, but
not on bars of greater weight, then LMH L 90 had 90 rein-

forcements, but LMH H 90 had only 30. Habit strength

20
prediction: The LMH L 90 group will make more responses
during extinction than LMH H 90, whereas as already predicted
no significant difference between L L 200 and H H 120 where
habit strength was asymptotic and equated. Furthermore,
if H H 120 makes more responses during extinction than
LMH H 90, this would indicate that an asymptote had not been
reached in LMH H 90 and that the conclusion of Capehart pp
pl. was open to question.

One important methodological consideration should be
mentioned, namely, the differential sensitivity of the bar
during extinction. The number of responses made in a Skinner
box is manually and/or electrically recorded, the criterion
for a response simply being a given excursion of the bar.
This allows for a clean, precise measurement of the number
of responses made, but when the bar is differentially
weighted, certain disadvantages are also apparent. An
animal extinguished on a light bar (5 8m.) makes many tan-
gential or incidental moves around the bar, which because
of the light bar sensitivity are recorded as responses,

i.e., the Operant level is high. This is not true when the
animal is extinguished on an 80 gm. loading which requires

a well-executed response to move the bar the prescribed
distance. One has only to observe the animals during
extinction to conclude that this is an important factor in
light-bar animals' showing a greater resistance to extinc-
tion. The incidence of exploratory behavior, which is minimal

after the first few acquisition sessions in the apparatus,

21
immediately increases during extinction following the first
flurry of responses. Much of this exploration consists
of the animal's standing on his hind legs, while smelling
and looking at the top of the box. Often, on returning to
a position on all fours, the animal bumps the bar, result-
ing in a recorded response on the light bar, but not on the
heavy bar. Since this is not an infrequent occurrence, the
tangential responses made by animals extinguished on a light
bar were recorded. They were then subtracted from the
total before a comparison was made with animals extinguished
on a heavy bar. Obviously, comparing two groups, when both
have been extinguished on a light bar, the subtraction is

not necessary.

III. RESULTS

The method of analysis throughout is Erratios of the
mean number of responses made during extinction by the
various groups for the specific comparisons dictated by
theory; two-tailed tests are used in every comparison.

Table 2, page 23, summarizes the data obtained from
Groups H H 120 and H H 200. This comparison was made first
since the two groups could be combined for a more reliable
comparison with Group L L 200 if they did not differ sig-
nificantly. The difference is not significant, indicating
that an asymptote had been reached at 120 responses.

The crucial comparison of animals trained and extin-
guished on a light bar with those trained and extinguished
on a heavy bar is presented in Table 3 on page 23. The
former group made more responses during extinction, and al-
though the difference is significant at the .05 level,
the difference is non-significant following subtraction
of tangential responses. The significant difference prior
to subtraction is discrepant with the findings of both
Maatsch, Adelman, and Denny (1954) and a portion of Aiken's
study (1957). Both these studies, as mentioned earlier,
found no differences between low and high effort groups in
extinction responding.

The mean number of tangential responses was 13.4 (155.7

- 142.3). However, it should be pointed out that this

22

23

Table 2.--A comparison of the mean number of responses made
during extinction by Groups H H 120 and H H 200.

 

Group M S.D. t df Level of
Significance
H H 120 110.1 46.7
.36 18 .40
H H 200 101.9 53-9

 

Table 3.--A comparison of the mean number of responses made
during extinction by Groups L L 200, and H H 120 and H H 200

combined.
W
Group M S.D. t df Level of
Significance

 

Before Tangential Bar Presses Are Subtracted

L L 200 155.7 56.2

2.67 32 .05
H H 120 plus 106.0 49.3
H H 200

After Tangential Bar Presses Are Subtracted

L L 200 142.3 53.8

1.95 32 .10
H H 120 plus 106.0 49.3
H H 200

 

24

number represents only those responses which were clearly
tangential; those that were questionable were not tallied.
The number of questionable responses greatly exceeded those
that were clearly tangential, and it could reasonably be
argued that the difference between L L 200 and H H 120 and
H H 200 combined would shrink to zero if it were possible
to identify all tangential responses. Nevertheless, we
must point out that at this point in our analysis the
evidence for a habit strength interpretation is quite incon-
clusive.

The comparison between Groups L L 200 and H L 120 pre-
sented in Table 4 on page 25 is the most damaging finding
of the present study to the concept of Ir. Since the re-
sponse effort required of each group during extinction is
identical, Ir would predict that the two groups should show
approximately equal resistance to extinction. But it can
be seen that Group H L 120 made almost Epipp the number of
responses made by Group L L 200. This difference is statis-
tically significant well beyond the .01 level.

Table 5, page 25, presents the data relevant to whether
the response facilitation in Group H L 120 was due to a
contrast effect or to the greater magnitude of the acquis-
ition_response. Here it can be seen that the difference
between HL L 120 and H L 120 is significant at the .01
level. Since the 30 light-bar responses inserted just prior

to extinction were enough to significantly reduce extinction

25

Table 4.--A comparison of the mean number of responses made
during extinction by Groups L L 200 and H L 120.

 

Significance

 

L L 200 155.7 56.2
6.55 22 <:.01
H L 120 310.8 57.9

 

Table 5.--A comparison of the mean number of responses made
during extinction by Groups H L 120 and HL L 120.

     

 

eredp”' ‘ “M _2. S.D. t df Level of
Significance
H L 120 310.8 57.9
3.25 151 <.01
HL L 120 193.0 104.0

 

1An F-test was significant, so degrees of freedom were
computed with Welch's formula.*

 

03
mm

*Welch's Formula for Degrees of Freedom:
2
S1

, )2

Of: -2

2 2 2
S 1 s 1
1 N1 + 1 N2 N2 + 1

+

2I
1'0

 

I‘m

2

26
responding, it appears that the facilitation in Group H L 120
was due to a contrast effect rather than the magnitude of
the response during acquisition.

To determine the extent to which the contrast effect
was removed by the manipulations used in HL L 120, this
group was compared with L L 200. The comparison is pre-
sented in Table 6 on page 27. Since the difference is non-
significant, it is possible to conclude that the contrast
was essentially eliminated by the insertion of 30 light bar
responses prior to extinction in the HL L 120 group.

Table 7, page 27, summarizes the comparison between
Groups LMH L 90 and LMH H 90, a partial replication of the
Capehart pp pl. study described earlier. The tangential
responses (6.6) are subtracted from the total number of
extinction responses for the light-bar group in the table.
It can be seen that the group extinguished on the light bar
(LMH L 90) made ppppp_pippp as many responses during extinc-
tion as the group extinguished on the heavy bar (LMH H 90).
The same ratio of heavy-to-light bar responses was found by
Capehart pp pl. The difference is significant beyond the A
.01 level even though tangential responses have been sub-
tracted from the light-bar group total. The original cal-
tention of unequal habit strengths in these two groups for
the training procedure used is clearly supported, for a
non-significant difference was obtained between L L 200

(tangential responses subtracted) and H H 120 and H H 200

27

Table 6.--A comparison of the mean number of responses made
during extinction by Groups L L 200 and HL L 120.

 

Group M S.D. t df Level of
Significance
L L 200 155.7 55.2 1
1.03 14 .20
HL L 120 193.0 104.0

 

1An F-test was significant, so degrees of freedom were
computed with Welch's formula.

Table 7.--A comparison of the mean number of responses made
during extinction by Groups LMH L 90 and LMH H 90.

 

Group M S.D. t df Level of
Significance
LMH L 90 97.4 36.4
4.88 18 (.01

LMH H 90 32.3 20.5

 

28
combined when it was clear an asymptote had been reached by
all groups.

That the habit strength in LMH H 90 for the heavy bar
was quite weak is apparent when this group is compared with
H H 120 (Table 8). Although the latter received only 30
more reinforcements, they made ppppp pippp_as many responses
during extinction (t = 4.37). The magnitude of this differ-
ence indicates that the assumption that rats trained on
light bars push heavier bars with equal facility is not

tenable.

Table 8.--A comparison of the mean number of responses made
during extinction by Groups H H 120 and LMH H 90.

   

 

Group M S.D. t df Level of
Significance
H H 120 106.0 49.3
4.82 141 <.01
LMH H 90 32.3 20.5

 

1An F-test was significant, so degrees of freedom were
computed by Welch's formula.

Table 9 on page 29 presents a summary of all comparisons

made in the present study.

 

llllll’l‘fllllll

29

Table 9.--Summary of comparisons made.

   

Group Treatment Mean Level of

 

 

Significance
H H 120 120 heavy-bar responses, 110.1
vs. extinguished heavy .40
H H 200 200 heavy-bar responses, 101.9
extinguished heavy
L L 200 200 light-bar responses. 155.7
vs. extinguished light
H H 120 without tangential bar presses '05
plus subtracted) 106.0

H H 200 120 and 200 heavy-bar re-
sponses, extinguished heavy

 

L L 200 200 light-bar responses, 142.3
vs. extinguished light

H H 120 (with tangential bar presses -10
plus subtracted) 106.0

H H 200 120 and 200 heavy-bar re-
sponses, extinguished heavy

 

 

L L 200 200 light-bar responses, 155.7
vs. extinguished light .01
H L 120 120 heavy-bar responses, 310.8
extinguished light
H L 120 120 heavy-bar responses, 310.8
vs. extinguished light .01
HL L 120 120 heavy-bar responses, 193.0

30 light-bar responses,
extinguished light

 

L L 200 200 light-bar responses, 155.7
vs. extinguished light .20
HL L 120 120 heavy-bar responses, 193.0

30 light-bar responses,
extinguished light

 

 

LMH L 90 30 responses on 5, 40, and 97.4
vs. 70 gm. bars; extinguished.lu¢m .01
LMH H 90 30 responses on 5, 40, and 32.3
70 gm. bars; extinguished heavy
H H 120 120 heavy-bar responses; 110.1
vs. extinguished heavy .01
LMH H 90 30 responses on 5, 40, and 32.3

70 gm. bars; extinguished heavy

 

IV. DISCUSSION

The findings justify a number of conclusions: (1)
There is no significant difference in extinction responding
between high and low effort groups when an asymptote has
been reached and operant level is considered; (2) Animals
trained on a heavy bar and extinguished on a light bar show
a marked facilitation; (3) The facilitation is the result
of a contrast in stimulation since the insertion of light-
bar reinforcements just prior to extinction essentially
eliminates it; and (4) The assumption that rats trained
on light bars depress heavier bars with equal facility is
rejected.

In general, the results indicate that the importance
of the effort variable has been over-emphasized in experi-
mental extinction, particularly when small amounts of work
are involved, such as in a Skinner box. It is important
to note that this conclusion is made for small amounts of
work, and says little or nothing about the effect of fatigue
upon behavior.

Since work or effort is of minimal importance it
follows that Ir is of minor importance in extinction. What,
then, accounts for extinction? One interpretation of the
present findings can be made in terms of a one-factor inter-

ference theory of experimental extinction, as incorporated

3O

31
in elicitation theory (Denny and Adelman, 1955). This is
particularly appropriate because this theory was the
impetus for the Maatsch, Adelman, and Denny study (1954),
which was the first to cast doubt on the importance of the
effort variable in extinction. Traditionally, three argu-
ments have been raised against an interference interpretation,
but none poses a threat to the elicitation position. The
first argument contends that it is not clear where the com-
peting response comes from. Elicitation theory holds that
the omission of reward is a frustrating situation which
elicits a class of avoidant responses which then compete
with the originally learned behavior. Two studies have
directly tested the validity of this assumption. Amsel and
Roussel (1952) employed a two-stage straight runway, with
each stage terminating in a goal box. Rats were run over
a long series of trials in which food was available in each
goal box, and then during extinction food was omitted from
the first box. The results showed a decrease in running
time for the second stage of the alley during extinction,
clearly indicating the energizing effect of frustration on
learned behavior.

Adelman and Maatsch (1956) examined more directly the
relationship between frustration and extinction. Using a
box 10 inches high, the response of jumping out of the box
was reinforced differently for three groups of rats. One
group was rewarded with "curiosity" satisfaction, and a

second group with food, the third group with escape from

32
frustration. This procedure resulted in great differences
in resistance to extinction. Group one extinguished very
quickly; group two at the end of about 60 trials; but
group three, rewarded with escape from frustration, showed
no signs of extinguishing at the end of 100 trials.

A second common objection to a competing-response
eXplanation of extinction has been that since extinction is
a non-reinforcement situation, there is nothing to reinforce
the competing response when it gppp appear. This argument
is not damaging to elicitation theory, since the theory
does not view reinforcement in a Thorndikian fashion, i.e.,
the strengthening of a closely preceding response. Rather,
reinforcement is viewed as occurring simultaneously with
the eliciting of the response, i.e., it ip the eliciting
of a response. Furthermore, the theory holds that the
omission of reward in an established behavior sequence con-
stitutes an instance of reinforcement (elicitation) the
same as the consistent elicitation of some response, or
class of responses, by food, water, shock, etc. Since the
strengthening of a particular response tendency results
from the consistent elicitation of that response, the con-
tinued elicitation of avoidant-type responses during extinc-
tion in a Skinner box strengthens these responses which
then compete with the original learned sequence.

The third argument against an interference theory of

extinction has centered around the fact that basically one

33

process, conditioning, is used to eXplain both acquisition
and extinction. This being the case, it is argued that
both should respond to experimental variables in the same
way, e.g., the spacing of practice. The contention is that
massing of trials results in rapid extinction while having
the Opposite effect on acquisition, but a number of weak-
nesses are apparent in this argument. First, a number of
studies have shown that massed trials do not necessarily
facilitate extinction. This indicates that variables other
than the massing alone are operating. Secondly, in those
studies in which massing has resulted in rapid extinction,
one can analyze the results in terms of degree of frustra-
tion. Indeed, one can argue that it is this variable rather
than the massing pp; pp which is related to resistance to
extinction, i.e., the greater the frustration, the greater
the incidence of competing responses, the more quickly
extinction occurs. The third very apparent weakness in the
argument is that it does not take into account the fact that
totally different responses are elicited during acquisition
and extinction. They are in fact antagonistic response
classes of approach and avoidance, so the expectation of
responding similarly to certain experimental variables seems
naive.

The rejection of the assumption that rats trained on
light bars depress heavier bars with equal facility also

supports a major contention of Denny and Adelman's elicitation

34

theory. The theory holds that ". . .the m nature of
the response elicited at any instant in time is a fundamental
consideration in explaining behavior" (1955. p. 4). Again,
one has only to observe the animal in a Skinner box to con-
clude that the depression of a light bar is ppp_the same
response as the depression of a heavy bar. During advanced
stages of acquisition training on a light bar, the rat often
positions himself with one paw on the food cup and the other
on the bar. This posture enables him to secure the food
pellet following a bar depression even before it comes to
rest in the cup. A heavy bar response is very dissimilar,
usually requiring the animal to lean over the bar and grasp
the back edge of it with both paws, his downward thrust
usually resulting in the rear legs rapidly sliding forward.
"Plowing a field" may suggest a certain class of responses,
but if one man plows riding a tractor and another following
a horse, the behavior involved is obviously different.

Additional evidence of the dissimilarity between heavy
and light bar responses and the importance of the exact
nature of the response is available in the findings of
Stanley and Aamodt (1954). They trained half of their gs
(rats) to depress a 50 gm. bar and the other half to depress
a 100 gm. bar. Half of each were then extinguished on a 50
gm. bar and half on a 100 gm. bar. The results indicated
a high positive correlation between the force requirement

during acquisition and the forcefulness of responding

35

during extinction. Specifically, §s trained on a 50 gm.
loading and extinguished on 100 gm. made significantly more
incomplete responses during extinction than gs extinguished
on loadings equal to, or less than, the loading employed
during acquisition. In other words, it appears that extinc-
tion responding was a function of the exact response which
was consistently elicited during training.

The problem of tangential and incidental responses
being recorded because of the sensitivity of a 5 gm. bar
was mentioned earlier. By utilizing an apparatus of a slightly
different nature in future research, such as in the Stanley
and Aamodt study (1954), the problem might be circumvented
or at least minimized. Rather than an all-or-none situation
in which a given excursion of the bar either closes or does
not close a microswitch, a pressure-sensitive bar might be
arranged so that any contact with the bar would be recorded.
If each bar contact were then translated into grams of
pressure on some continuous recording device, the advantage
would be obvious. Responses made during heavy-bar extinction
which equalled or exceeded the amount of pressure necessary
to activate the light bar could then be added to the total
number of responses made. This would be a more objective
procedure than subtracting what appear to be tangential
responses made by animals on a light bar, as in the present
study.

The facilitation observed when animals were trained

on a heavy bar and extinguished on a light bar is of

36

interest in terms of the generalization-decrement hypothesis.
Any extinction procedure involves a change from the acquis-
ition situation in that the proprioceptive consequences of
reinforcement, and eventually responding, are eliminated.
Generalization decrement extends this idea and postulates
that the speed of extinction is inversely related to the
similarity in stimulation between acquisition and extinction.
One might argue that the theory would make a wrong predic-
tion in the present study since it would hold that §3 trained
on a heavy bar and extinguished on a light bar would extin-
guish faster than gs trained on a light bar and extinguished
on a light bar. The prediction would be based on the fact
that kinesthetic feedback differs during acquisition and
extinction in the former but not in the latter. However,
this argument includes the assumption that the animals :22;
the difference in stimulation--which may or may not be true.
It seems more apprOpriate to question the applicability
of the generalization-decrement hypothesis in the present
case rather than to reject it outright.

Differing kinesthetic feedback was used by Aiken (1957)
. in an attempt to account for the fact that his high effort
gs showed the same resistance to extinction as low effort gs,
during the first extinction session. He suggests that the
feedback stimulation from the bar-pressing responses is
analogous to Hull's V (stimulus intensity dynamism), and
that the prediction of a greater habit strength following

high rather than low effort can be obtained from the formula:

37
SHr = SHr x V1. The present findings, however, indicate that
this explanation is inappropriate. The facilitation result-
ing from heavy bar training and light bar extinction was
essentially eliminated by inserting light bar reinforcements
prior to extinction; therefore, the increment in extinction
responding is a function of a stimulus change and ppp
greater habit strength following high effort. This in effect
destroys Aiken's attempt to make his data mesh with Hullian
theory.

At this point one might speculate as to why Aiken's §s
which were trained light and extinguished light made more
responses than gs trained heavy and extinguished heavy
during the second extinction session, but not during the
first. The panel-push response employed is no different
from a bar-press response in that tangential responses would
augment the low-effort group total but not the high effort
total. If it can be assumed that the incidence of explor-
atory behavior is greater during the second extinction
session than during the first, it follows that the number
of tangential responses would also be greater and might
account for the difference between the two groups. This
interpretation seems especially plausible since any movement
by the animal which displaced a swinging door 1/2 inch
within 30 seconds after the door was exposed counted as a
response and terminated the trial. The dependent variable

was defined as the number of trials to extinction, and

38
extinction was said to have been reached when § failed to
respond on two consecutive 30 second periods with the
swinging door exposed.

Applezweig's (1951) conclusion was the same as Aiken's.
He found a significant positive relationship between effort
during training and number of extinction responses. This
was largely a result of the facilitation of Se trained heavy
and extinguished light because he reports that when animals
were trained and extinguished at the same effort, there were
not ". . .any determining factors, other than chance, under-
lying the correlation between effort level and rate of
extinction" (1951). p. 230). He then concludes that there
is greater habit strength when the effort involved in
learning is greater, which according to the present results
is erroneous.

Facilitation as a result of a contrast in stimulation
is not a new phenomenon, nor is it restricted to a particular
set of experimental conditions. If one views it as the
instrumental analogue of positive induction in classical
conditioning, the first mention of it was made by Pavlov
(1927). Skinner (1938) spoke of a contrast effect in his
book, and Verplanck (1942) and Solomon (1943) observed it
in running speed in a maze and in jump-stand latency,
respectively. It has frequently been observed in studies

of incentive magnitude when ps are switched from one incentive

value to another (Crespi. 1942). Aiken (1957) referred to

39
it in his instrumental panel-push study. One of the most

recent examples (Brethower and Reynolds, 1962) demonstrates
the facilitative effect of punishment on unpunished behavior.
The experimenters reinforced the key pecking of pigeons on

a variable-interval schedule during the presentation of each
of two stimuli (red and green lights). Later, punishment
followed every response emitted in the presence of the red
light, but not in the presence of the green. The result

was that when the rate of punished responding changed during
the presentation of the one stimulus, the rate of unpun-
ished responding during the other stimulus changed in the
opposite direction. In other words, when punished respond-
ing decreased, unpunished responding increased; and when
previously punished responding increased following termin-
ation of the shock, the always unpunished responding de-
creased.

It is of interest to speculate as to why such a pro-
nounced facilitation occurred in the present study, while in
some other studies it was absent or greatly diminished. One
explanation might be that two essential conditions for its
appearance involve asymptotic habit strength and a large
differential in the light and heavy bar loadings, both of
which are characteristic of the present study, but missing
in some previous works. For example, the loading differ-
ential in the Capehart pp_pl. study (1958) was 65 gm.

(70 gm. minus 5 gm.) but just prior to extinction the S3

40
were reinforced on a 40 gm. bar, not a 5 gm. bar. Also,
these data strongly suggest that an asymptote had not been
reached. In the Stanley and Aamodt (1954) experiment, an
asymptote was established (124 acquisition responses at the
extinction bar loading) but the weight differential was only
50 gm. (100 gm. minus 50 gm.). The present study included
the establishment of an asymptote and a differential of

75 gm. (80 gm. minus 5 gm.).

Ipplications for Future Reseapgh

In general, little is known about facilitation result-
ing from contrast effects, particularly concerning the
parameters which affect it. Pereboom (1957) has, in part,
accounted for the phenomenon in incentive magnitude studies
with an explanation based on competing, exploratory behavior,
but other areas remain in need of research. An immediate
question suggested by the present study concerns the manipu-
lation required to eliminate the demonstrated contrast
effect. Since following heavy—bar training, 30 light-bar
reinforcements essentially eliminated the facilitation,
one might ask what are the minimum number of reinforcements
necessary to achieve the same result. It is possible that
the number is related to the length of time at the onset
of extinction that an animal continues to make a heavy-bar
response when trained on a heavy bar but extinguished on
a light bar. That is, following heavy-bar training, an
animal extinguished on a light bar initially exerts enough

41
force to depress an 80 gm. bar when a 5 gm. effort is
sufficient to activate the food delivery mechanism. This
excessive force in responding usually diminishes anywhere
between the tenth and fiftieth extinction response, and, as
in facilitory effects, perhaps represents the minimal amount
of kinesthetic feedback necessary to effect a behavior

change.

V. SUMMARY

This study was designed to determine the importance
of effort as a variable in experimental extinction when the
habit strength of the response has been controlled. Seven
groups of animals were given various amounts of training
on differentially weighted bars in a Skinner box with the
following results: a

1. Contrary to the prediction of Ir, the difference
in resistance to extinction between groups trained and
extinguished on a light bar and those trained and extin-
guished on a heavy bar is non-significant when an asymptote
has been reached and operant level is taken into account.

2. A partial replication of a previous study in
which effort and extinction were shown to be directly related
indicated that an asymptote had not been established, and
that the findings are better interpreted in terms of differ-
ential habit strengths.

3. Also contrary to the prediction of Ir, animals
trained on a heavy bar and extinguished on a light bar show
a facilitation when compared with animals trained and extin-
guished on a light bar.

4. The facilitation resulting from heavy bar training
and light bar extinction was significantly reduced by
inserting light bar responses prior to extinction. This
suggests that the facilitation is a function of a contrast

42

43
effect and not the magnitude of the acquisition task. The
incidence of contrast effects in various experimental
situations was briefly reviewed and the lack of informa-
tion of the parameters involved was pointed out.
5. An interference theory of extinction was suggested
as being more parsimonious and appropriate in accounting

for the present findings.

10.

11.

REFERENCES

Adelman, H. M., and Maatsch, J. L. (1956) Learning
and extinction based upon frustration, food reward,
and exploratory tendency. J. exp. Psychol., 52, 311-315.

Aiken, E. G. (1957) The effort variable in the
acquisition, extinction, and spontaneous recovery of
an instrumental response. J. exp. Psychol., 53, 47-51.

Amsel, A. and Roussel, J. (1952) Motivational proper-
ties of frustration: I. Effect on a running response

of the addition of frustration to the motivational
complex. J. eXp. Psychol., 43, 363-368.

Applezweig, M. H. (1951) Response potential as a

function of effort. J. comp. physiol. Psychol., 44,
225-2350

Brethower, D. M. and Reynolds, G. S. (1962) A facili-
tative effect of punishment on unpunished behavior.
Jo 8X9. Anal. BehaVO9 5’ 191-1990

Capehart, J., Viney, W. and Hulicka, I. (1958) The
effect of effort upon extinction. J. comp. physiol.
Psychol., 51, 505-507.

Crespi, L. P. (1942) Quantitative variation of
incentive and performance in the white rat. Amer. J.
PSYChOlo, 55' 467-5150

Denny, M. R. and Adelman, H. M. (1955) Elicitation
theory: II. The formal theory and its application to
instrumental escape and avoidance conditioning.
Unpublished paper, Department of Psychology, Michigan
State University.

Glanzer, M. (1953) The role of stimulus satiation
in spontaneous alternation. J. exp. Psychg;., 45,

387-3930

Hull, C. L. (1943) Principles of behavior. New York:
D. Appleton Century.

Jaynes, J. (1950) Learning a second response to a
cue as a function of the magnitude of the first. i.

comp. physiol. Psychol., 43, 398-408.

44

i.l|lllil.lllllll"i[i

 

[I ll‘l 4! i
I! II I

45

12. Kendler, H.H. (1962) Basic psychology. New York:
Appleton-Century-Crofts, Inc.

13. Kimble, G. A. (1961) Hilgard and Marguis' Conditioning
and Learning. New York: Appleton-Century-Crofts, Inc.

14. Lewis, D. J. (1963) Scientific principles of psychologg.
New Jersey: Prentice-Hall.

15. Maatsch, J. L. (1955) An experimental test of the
differential effects of work and frustration upon
learning. Unpublished Ph.D. thesis, Department of
Psychology, Michigan State University.

16. Maatsch, J. L., Adelman, M., and Denny, M. R. (1954)
Effort and resistance to extinction of the bar pressing
response. J. comp. physiolp Psy§h01., 47, 47-50.

17. Montgomery, K. C. (1952) A test of two explanations

of spontaneous alternation. J. comp. physiol. Psychol.,
45. 287-293.

18. Mowrer, O. H. and Jones, H. M. (1953) Extinction
and behavior variability as functions of effortfulness
of task. J. exp. Psychol., 33, 369-386.

19. Pavlov, I. P. (1927) Conditioned reflexes. (Trans.
by G.V. Anrep) London: Oxford University Press.

20. Pereboom, A. C. (1957) A note on the Crespi effect.
Psyphol. Rev., 64, 263-264.

21. Skinner, B. F. (1938) The behavior of organisms.
New York: D. Appleton Century Co.

22. Solomon, R. L. (1943) Latency of response as a
measure of learning in a "single-door" discrimination.
Amer. J. Psychgl., 56, 422-432.

23. Solomon, R. L. (1948a) Effort and extinction rate:
A confirmation. J. comp. physiol. Psychol., 41, 93-101.

24. Solomon, R. L. (1948b) The influence of work on
behavior. Psychol. Bull., 45, 1-40.

25. Stanley, W. C. and Aamodt, M. S. (1954) Force of
responding during extinction as a function of force
requirement during conditioning. J. comp. physigl,
Psychol., 47, 462-464.

26. Verplanck, W. S. (1942) The development of a discrim—
ination in a simple locomotor habit. J. exp. Psychql.,

27.

28.

46

Walker, E. L. (1956) The duration and course of the
reaction decrement and the influence of reward. J. comp.

physiol. Psychol., 49. 167-176.

Walker, E. L., Dember, W. N., Earl, R. W., Fawl, C. L.,
and Karoly, A. J. (1955) Choice alternation: III.
Response intensity vs. response discriminability.

J. comp. physiol. Psychol., 48, 80-85.

l‘|5l‘5'|'lllnllll'll'Ill‘illiiil’llllll

v’t 1!. ‘-‘ ~"‘ “‘4?“ .
131133.72" '1‘??- 1132‘!
Hump-1 31.1!th uni.

 

4

1293 03177 60

Ill
"1.
"II
III
E||
"
H
H

3