, ..
-.,-.. ....
w'

I"

I’I‘III ’I  " I
III I IIII I

I

I
I I

I I
III
I I

I
I
l

I

I |‘ “I

II
I II

I I 3
II ‘II

I
I

I
I

I
I
.
II I,

I
I
I
I

__\
—\
\I

 

AN EXPERIMENTAL STUDY OF THE
EFFECT OF THE INITIAL REINFORCEMENT
ON SUBSEQUENT RESPONSE
TENDENCY UNDER DIFFERENTIAL

CONDITIONS OF FOOD EXPECTANCY

Thesis for the Degree of M. A.
MICSIIGAI’J STATE CfJLLEGE
Robert Lloyd MarfindaIe
1950

This is to certify that the

thesis entitled

"An uperlnental etudy of the effect of the initial
reinforcement on the subsequent tendency under
differential conditions of food expectancy." \

presented by

Mr. Robert Hertindale

has been accepted towards fulfillment
of the requirements for I

4L;— degree in W

2%an

Maflr professor

Date “3’ 259 1.21—

0-169

 

.--.m Q... ~ __...

7-,- __..._

 

 

 

 

 

 

.
' ' v- v r __I_——.
) I. , AI? .._'W|\ - n—
“ ‘ u v3..n.'v;.—'_._. -

AN EXPERIMENTAL STUDY OF THE EFFECT OF THE INITIAL
REINFORCEMENT ON SUBSEQUENT RESPONSE TENDENCY
UNDER DIFFERENTIAL CONDITIONS OF
FOOD EXPECTANCY

by
Robert Lloyd Martindale

A THESIS
Submitted to the School of Graduate Studies of Michigan
State College of Agriculture and Applied Science
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Psychology

1950

THESIS

 

 

 

 

ACKNOWLEDGEMENT

Grateful acknowledgement is made to Dr. M. Ray
Denny, under whose supervision this study was carried
out, for his guidance and participation in the work
herein reported. The authbr also extends his _
gratitude to Dr. Lee Katz for outlining the complex
statistical procedures and to Richard.L. Behan for
his indispensable aid in carrying out the statistical

analyses.

f2f1(lflflfi

TABLE OF CONTENTS

LIST OF TABLES 0.00.00.00.00.00000000000000...
LIST OF FIGIIRE 00....OOOOOOOOOOOOOOOOOOOOOOOO

I.
II.
III.
IV.

VI.

VII.

INTRODUCTION OOOOOOOOOOOOOOOOOOOOOO.
THEORETICAL ORIENTATION ............
EDG’ERIMENTAI. HYPOTHESES ............
EflmmMMmemmEnunuun.
Experimental Apparatus

Subjects

Habituation Period

Preliminary Training

Training

Extinction

RESULTS AND DISCUSSION .............
SUB'EEEIIIMENT 00.000000000000000...

Procedure
Results and Discussion

SUMMARY AND CONCLUSIONS ............

BIBLIOGRAPHY 0000.00.00.00...00000000000000000
APPMIX 00000000000000.0000.0.000000000000000

11

Page
,111
iii

14
15

26
39

44

46
47

Table

I.

II.

III.

V.

Figure
~Experimental Apparatus Floor Plan ........

I.
II.

III.

LIST OF TABLES

Analysis of Variance of Training Data for
Preliminary Training Groups and Training
Groups in Terms of the Number of Correct
Responses per Animal .....................

Group Comparisons in Terms of the Mean
Number of Correct Responses for the Total
Training Perioa 0000000000000eeeeoeeeeoeoo

Comparison of Preliminary Training Groups
and Training Groups in Terms of the
Number of Correct Responses on Pertinent
Training Trials eeeoeeeeeeeeeeeoooeooooeee

Analysis of Variance of Extinction Data
for Preliminary Training Groups in Terms
of the Number of Correct Responses per

Animal eeeeeeoeoooeeoooeoooeeoooooo.ooeeoo

Analysis of Variance of Eating Latency
Data for Preliminary Training Conditions
and Training Conditions in Terms of the
Mean Latency per Day per Animal ..........

LIST OF FIGURES

Learning Curves of Composite Groups I, II,
and III in Terms of Per Cent of Ultimate
Correct Responses on Each Trial ..........

Latency of Eating Response Curves for
Composite Groups A and C in Terms of the
Mean Time Per Trial Per Day 0000000000000.

111

Page

27

28

29

33

34

Page

16

31

35

eyenooeoeoeeeeuOOIQQUDIOO‘

..-I-e.geecoco-con'eeeo'euevglfiu

I I O O O ‘ O -

Cote-cult.

I INTRODUCTION

The theoretical background of the present study
is oriented around basic philosOphic concepts most
generally referred to as scientific mechanism or
determinism. It allows no consideration of mystical
or metaphysical concepts or of any viewpoint that
postulates anything that is not ultimately observable
by the scientific community. All constructs (mechanisms
of observable relationships) must be defined in terms
of the observable either directly or indirectly. A
similar viewpoint is expressed by Hull as follows:

whereas argument reaches belief in its

theorems because of antecedent belief

in its postulates, scientific theory

reaches belief in its postulates to a

considerable extent through direct or

observational evidence g£_the soundness
25 its theorems. (1)

 

 

(1) Hull, Clark L., Principles of Behavior, p. 9, .
1945. This work will hereafter be referred to as P. B.

 

A living organism is defined as any relatively
unitary collection of matter that is capable of
differential adaption to an external stimulus
situation (physical energy external to the organism)

or internal stimulus situation (physical energy

(1)

components of a need state necessary to the maintenance
of the organism in its essential unitary state). -In
order to meet this requirement the living organism is
essentially a communications or feedback system in the
process of a single continuous Operation that transmits
physical stimulus energies. The nervous system with its
receptor beginning and its effector termination is in
essence just this sort of communications mechanism.

This viewpoint of the living organism is derived almost
totally from the implications of the field of

"cybernetics".(2) The organism can neither add nor

 

(2) Wiener, Norbert, beernetics, 1948.

 

subtract from these stimulus energies but only receives,
reintegrates, and distributes them in a manner most
expedient to its maintenance in so far as these energies
are within the reception and dissemination range of the
organism. This point of view bears some similarity to
Hull's conception of organisms as "self-maintaining

mechanisms" (5) but our viewpoint is elaborated on a

 

(3) Po Bo, p.384.

 

more molecular level.
From this it is assumed that the organism is a

stimulus-response mechanism. On a molar level such a

mechanism consists of: (l) a complex physical structure
for communications, and (2) a totality of all stimulus
energies that have impinged on the receptors and have
not been dissipated by the effectors. Since the
subject matter we are dealing with is the molar,
observable components of the organism's adjustment to
its environment, we can postulate theoretical constructs
that allow us to predict lawful relationships between
afferent and efferent aspects of the maintenance of
the organism.

Watson conceived of an identity between the
stimulus and the response such that either could be

predicted from the other.(4) This identity hypothesis

 

(4) Watson, John B., The Ways 2; Behaviorism, p. 2,
1928.

 

is essentially congruent with the presented point of
view but Watson minimized the obvious complexities of
the relationship. However, Hull, although he implicitly
accepts this identity hypothesis, further elaborates the
necessity for the postulation of the intervening
constructs logically deduced from the observable, molar

antecedents and consequents.(5)

(5) Hull, Clark L., "Mind, Mechanism, and Adaptive
Behavior", Psychological Review, vol. 44 (1937), p. 1-32.

 

II THEORETICAL ORIENTATION

Learning of an instrumental act is dependent upon
the establishment of bonds or relations between an
instrumental act and a stimulus complex in a rewarding
situation. A rewarding situation consists of: (1)
reduction of a physiological need that is based on the
innate structure of the organism, or (2) a stimulus
situation that has been in temporal contiguity with
physiological need reduction and therefore represents
actual need reduction. This viewpoint of reinforcement
is almost identical with Hull's latest revision of his
postulate and corollary concerned with primary and

secondary reinforcement.(6)

 

(6) Hull, Clark L., Behavior Postulates and Corollaries
‘; November, 1949, p. 2.

 

Denny suggests that all reinforcement of an
instrumental act is, in Hullian terms, secondary
reinforcement. Because actual need reduction does not
take place in sufficiently close temporal contiguity with
an instrumental act to establish learning bonds,
secondary reinforcement is what is actually present.

This would be particularly true at the beginning of
training. Denny also postulates a view of reward similar

to the above. He classifies reward in terms of the

(4)

demands of the organism's permanent or innate structure
and temporagy structure (momentary state of the organism).
These states predispose the organism toward a particular
response, and when such a response occurs a reinforcing
state of affairs exists. Fractional-anticipatory-

responses are included in the latter category.(7)

 

(7) Denny, M. Ray, Unpublished lecture series (untitled),
1950.

 

It is implicit in the preceeding postulations that
all rewarding situations must occur within the temporal
duration of a superthreshold neural stimulus trace in
order for that stimulus to get hooked up to the
rewarding situation. The shorter the time between the
onset of a neutral afferent neural impulse and the
onset of an afferent neural impulse of a rewarding
stimulus situation, the stronger will be the learning
bond thus established. Therefore,‘i£‘a neutral
instrumental £33 occurs'ip sufficiently glpgp temporal
contiguitywpiph'g rewarding stimulus situation, i3 ggp‘pp
postulated that learning pf £22 instrumental act gill

occur with the first presentation pf the rewarding
stimulus situatIUuT'

Hull presents a similar viewpoint to this in his

"law‘pf primary reinforcement" without the specific

reference to the effect of the first reinforcement as

follows:

Whenever an effector activity occurs in
teppo oral contiguity with the afferent—
impulseE Lr the perseverative trace 23'such

impu se, resulting from the impact L
a —st1mulus energy upon_ a rec_ptor, and—
this conjunction is closely associated in
time with the d minuation in the receptor

scharge characteristic ofa need there
will result an increment EE’Lhe endenc
for that stimulus Ln subsequent occasions
to evoke that reaction. (8)

 

 

(8) P. B., p. 80.

 

A similar interpretation is also inherent in Hull's
postulation of the "gradient of reinforcement" as
different from the "goal gradient". The "gradient of
reinforcement" represents an interval of time after an
instrumental response in which secondary reinforcement
is not necessary for an increment in that response
tendency. After this primary temporal duration secondary
reinforcement of some type is necessary to produce an

increment.(9) This postulation seems to imply that what

 

(9) P. B., p. 158.

 

we are dealing with in the case of the “gradient of
reinforcement“ is a perseverative stimulus trace.

The experimental evidence from.which Hull measures
the "gradient of reinforcement" has been attacked by
Spence and Grice on the grounds that all secondary
reinforcement had not been eliminated during the time

delay and for that matter could not be eliminated

because of the secondary reward value of the instrumental
act itself. Spence states his position as follows:

The interpretation that learning under
conditions of delay of primary reward
involves a backward action of the goal
object on the preceeding stimulus-response
event is rejected. The hypothesis suggested
as an alternative to this conception is

that all such learning occurs as a result

of the develOpment of secondary reinforcement,
the action of which is conceived to take
place immediately upon occurance of the
response.(10)

 

(10) Spence, K. W., "The Role of Secondary Reinforcement
in Delayed Reward Learning" , Psychological Review, vol.
54 ‘1947), p. 1-80

 

Grice concludes from.experimental results that when all
secondary reward is eliminated subsequent to the
instrumental act, learning may still occur within a
substantial time interval. He states further that "as
long as the trace of the correct response is within the
range of the generalization gradient of the prOprioceptive
pattern stimulating the organism at the time of the
response in the (correct) alley, the prOprioceptive

pattern will acquire secondary reinforcing prOperties".(11)

 

(ll) Grice, G. Robert, "The Relation of Secondary
Reinforcement to Delayed Reward in Visual Discrimination
Learning" , Journal of Experimental Psychology, vol. 38

(1948), p. 15.

 

The author and Denny have both interpreted these

formulations as meaning that secondary reinforcement

value of the instrumental response is a necessary
condition to learning when all other forms of secondary
reward have been eliminated during the delay period.
However, before the first reinforcement the instrumental
act has no reward value. Therefore, if there is an
increment to response tendency immediately after the
first reinforcement, it must be due to a primary
"gradient of reinforcement". The goal gradient cannot
be Operative at this point because the instrumental
response could never have been experienced in absolute
contiguity with the secondary reward of the goal
gradient until after the second reinforcement.

In a more recent revision of his postulates Hull
explicitly postulates the nature of the stimulus trace
as a function of intensity and time. The action of the
stimulus in the afferent nervous system after its
physical termination can still serve as a learning
stimulus when brought into contiguity with primary or
secondary need reduction. Although, again, there is no
specific reference to the first reinforcement, it would
seem inconsistent, if the stimulus trace of the
instrumental response contiguous with the first
reinforcement of that response did not result in an
increment to the tendency for that response to be evoked

upon a future presentation of those stimuli.(12)

 

(12) Hull, Clark L., Behavior Postulates and Corollaries
; November, 1949, p. 2.

 

The author regards behavioral oscillation as the
behavioral component of a physiological need state and
as such anticipatory response is not believed to
mediate learning on a molar level. In as much as the
physiological need state of the organism.may vary from
moment to moment it follows, if oscillatory behavior
is correlated with the need state, that this behavior
will vary from moment to moment. In other words
continuous change in the structure of the organism is
presumably the basis of behavioral variation or
oscillation. Thus we have a need state basis for
oscillation rather than attributing oscillation to a

"little understood physiological process" as Hull

does.(13)

 

(15) P0 B., p. 3930

 

Exploratory responses as behavioral manifestations
of oscillation would necessarily have both external and
internal stimulus correlates. Therefore, it would seem
on the basis of former postulations that exploratory
stimulus components because of their close temporal
contiguity with the rewarding situation would gain

secondary reward value. However, if we examine this a

10

little closer we see that such is not the case. It is
true that the close temporal contiguity of these
stimuli to the rewarding situation would provide
Optimum conditions for learning, but these stimuli and
their responses are present at other times when they
are not rewarded and, therefore, extinguish. In fact,
it is obvious that these responses are unrewarded in
most instances of their occurrence,as for example, with
a hungry animal in an empty home cage. In view of this
interpretation it appears highly doubtful that there is
any learning of an exploratory response on a molar level.

Actually Hull's interpretation does not differ
from this to any great extent. The primary need state
initiates and maintains anticipatory responses.
Although he states that the resulting stimuli would
establish learning bonds to rewarding situations, he
does not regard them as an important factor in the

learning of a response sequence.(l4) However, it should

 

(14) Hull, Clark L., "Goal Attraction and Directing
Ideas as Habit Phenomenon", Psychological Review, vol.
38 (1931). p. 504.

 

be further noted that Hull regards the "fractional
component of the goal response" as something quite
different from the anticipatory response. This
"fractional component of the need reduction process"

is regarded only as a possible mechanism.of reinforcement

ll

generalizing back along the goal gradient.(15)

 

(15) P. B., p. 100.

 

It is postulated that behavior is specified in
terms of the 2229.32222 by the need state itself. It
specifies behavior to the extent that non-rewarded
exploratory responses are more quickly inhibited and
rewarded responses are more quickly facilitated. The
greater the deprivation of an organism in terms of
the particular physiological need, the less will be
the variability of behavior in a rewarding stimulus
situation. Therefore, response latency is postulated as
a negatively correlated function of the degree of
deprivation, and is not a direct measure of habit
strength. The degree of deprivation does not facilitate
learning directly but only provides more Optimum
conditions for learning by decreasing the time interval
between response and reward. With the degree of
deprivation varied and the time interval between an
instrumental act and reward held constant, there should
be no difference in the probability of occurrence of the
instrumental response.

From the writer's point Of view behavior is
specified in terms of the particular learning situation
by a successive spatial movement of the rewarding

stimulus situations back to previously neutral stimuli

12

until the instrumental act is reached. It should be
noted that this is a molar interpretation and that all
stimuli in the experimental situation would establish
learning bonds to the rewarding situation at least on a
subliminal level in terms of the perseverative stimulus
trace. This process continues infinitely back from the
initial rewarding situation so that behavior becomes
specified in terms of more remote and more general
stimuli on a more and more molar level.

Therefore if an animal is rewarded in a stimulus

 

context spatially antecedent 22 fig instrumental 333‘
then the animal's future learning RE the instrumental
3g; should £23.22 facilitated. An Opposing viewpoint
would be that if an animal is rewarded previous to

the instrumental act, then fractional-anticipatory-
responses or expectancies would be set up to the
stimulus pattern. These expectancies would specify
behavior in terms of the relative need state and this
specification Of behavior would facilitate learning of

a subsequent instrumental response. FOr example Dorothy
N. Moore in a Master's thesis concludes on the basis of
her experimental results that "in the absence of
expectancy on the first rewarded trial, no learning of
the instrumental act takes place; therefore an expectancy

or anticipatory set seems essential for instrumental

response learning".

13

This conclusion appears to be in antithesis to the

present theoretical framework.(16) However, an

 

(16) Moore, Dorothy Nell, 5g Experimental Study of the
Effect of Position Reversal after One or Two Reinforcements
33 SimfiIE‘T:Maze Learningf:n_th§_Rat 'Thesis Tbr the

Degree _O_f g; _A_._ (Michigan state College, 1949TT-p779.

 

 

 

 

examination of her experimental apparatus reveals that

the instrumental act was temporally so distant from the
rewarding stimulus situation that the stimulus trace of
the instrumental act was probably subliminal at the time

of reward.(17) In this case, therefore, no obServable-

 

(17) Ibid., p. 15.

 

evidence of learning would have been detected after the

first reinforcement.

III EXPERIMENTAL HYPOTHESES

The previous postulations assume that expectancies
do not mediate learning. Therefore, the following

hypotheses are evident in the experimental design:

A. If, in a T-maze learning situation designed so as to
reduce the time interval between the instrumental act
and the reinforcement to minimum, the end boxes associated
with reward and non-reward are reversed after one
reinforcement, the animals' learning of the instrumental
act should be retarded as compared with the animals'
having no reversal.

B. If, in the same T-maze situation, the end boxes are
reversed after the second reinforcement, the animals'
learning of the instrumental act should be retarded as
compared with animals having no reversal and this
retardation should be greater than in hypothesis A.

C. If animals are trained so as to set up food
expectancies to the starting box and straight alley,
their subsequent learning of the instrumental act

should not be in any way facilitated as compared with
naive animals as long as both have approximately
immediate secondary reinforcement in the subsequent

learning situation.

IV EXPERIMENTAL PROCEDURE

A. Experimental Apparatus

The apparatus was a modified, enclosed T maze (see
floor plan in Fig. 1). It was specifically designed to
make the delay of reward of the instrumental act as
short as possible. The response from the choice point
was a continuously similar turning response either to
the right or the left. The entire apparatus was covered
with a-} in. hardware cloth except the starting box
which had an Opaque wood cover. The floors were 3/4
in., the sides % in., and the doors % in. plywood. The
curved surfaces of the maze were constructed of heavy
gauge galvanized metal. The inside depth of all parts
of the maze was 8 in. All interior surfaces of the maze
except the goal boxes were painted with a neutral gray
enamel. The entire apparatus was placed on a 43% in.
by 22% in. table which was 27 in. high.

The doors at the choice point were Operated by a
release mechanism that could drOp them swiftly when the
first part Of the rat's body crossed the response
criterion line. At this point the door from the straight
alley and the door farthest from the end box that was
entered were closed. As soon as the animal had completely

entered an end box, the door nearest to the end box

(16)

 

l6

 

 

 

B - Starting Box
- Stem and
traiaht Alley

 

.4,

r-----«

U)

 

.._.........._._.————_'L
pin-m2

SB

 

 

 

 

”in-..“

Figure T.

q-‘—‘ , _.. . .. _.
x; ._ ‘2’ ‘nT .'. ‘
A- I. .5- I LA. ... 1... .4

1/ 4 122m:

- N21). gal}
- Posit

 

V" .771 22

-. I ‘ 3—. W};“. ‘r h .
.~. .1...Li~ i4 3.2.1“.

m

OT f
2.1.3. .

f

' .;~:.2 12122.12

L

17

entered was closed. The door at the starting box was
equipped with a conventional string, pulley, and weight
arrangement. All doors were padded with sponge rubber
at the bottom so that their closing was as silent as
possible.

Forced trials were performed by placing a false
wall Of % in. plywood on the opposite side Of the
choice point from the maze arm into which the animal was
to be forced. There was one of these for each side.
They fitted flush against the rear wall on either side
of the choice point and against the curved surface ‘
leading to the maze arm from the choice point in such
a way that the maze arm was completely blocked Off,
forming a solid straight wall extending from the starting
box to the Back of the choice point on the desired side.
The false wall was the same height as the rest of the
maze. It was painted with the same gray enamel as the
rest of the maze interior.

The end boxes were of distinctly different shape
as illustrated and were the same height as the alleys.
The negative and box was lined on the floors and sides
with white posterboard. The posterboard formed a
false wall with a % in. space between the posterboard
and wooden walls of the box. This space was filled with

crushed food identical to that used in the positive goal

18

box in order to equalize the food odor. The positive
goal box was lined with black posterboard on the walls
and floor. The entire floor of the positive goal box
was covered with Purina Dog Chowtlheckers of the same
type as the animals were normally fed. The posterboard
was used rather than paint in the goal boxes in order to
minimize any differential paint odor.

The goal box for preliminary training was placed
at the end of the straight alley. It was constructed of
plywood with inside dimensions of ll 5/8 in. wide, 7% in.
deep, and 5% in. high with a rectangular floor plan.

It was covered with a hinged cover of ordinary window
screening. The entire interior of the box was painted
with a neutral gray, flat paint. About midway through
the experiment this box was replaced with another
rectangular box with inside dimensions of 5 in. wide,
12 in. deep, and llfi-in. high. It was covered with a
cover made of i in. mesh hardware cloth. The interior
was painted with a neutral gray, flat paint.

The maze was lighted by a clear 200 watt bulb
placed directly over the center of the straight alley.
The bulb was 68 in. from the floor of the alley. There
were also 60 watt frosted bulbs in semi-spherical white
reflectors shining directly into each goal box and
respective maze arm. The bulbs were placed 4 in. from

the top of the walls of the maze. Both of these lamps

19

were placed above the first training goal box 6 in. from
the top of the walls of the box and above the second
training goal box 1 in. from the tOp of the walls of

the box. The lighting procedure above reduced
differential illumination, resulting from the reflection
of‘the white and black walls of the respective end boxes
on the maze arm walls that was visible from the choice
point, so that it was just barely discriminable to the

experimenter.

B. Subjects

The subjects were experimentally naive albino rats
from the rat colony of the psychological laboratory of
Michigan State College. The animals varied in age from
60 to 165 days at the beginning of the experiment.
Their mean age was approximately 110 days. The animals
remained in their typical home environment throughout
the experimental period, except that they were fed
separately in individual feeding cages for part of the
period. The animals were fed Purina Dog Chow Checkers
both in the experimental apparatus and in the feeding
cages. This was the same food that they were normally

fed in their home environment.

20
C. Habituation Period

All animals except for four in the first group,
who are designated in the data, were given a two day
habituation period. This consisted of handling by the
experimenter and being allowed to run free on a small
table. On the first day of this period the animals were
marked by a combination of cuts in their ears and ink
markings on their tails. On the first day of this
period they were given a large amount of food in their
home cage. They were not fed on the second day. Water
was continuously available both of these days and

throughout the remainder of the eXperimental period.

D. Preliminary Training

On the day immediately following the habituation
period a five day preliminary training period was begun.
The animals were divided into two training groups at the
beginning of this period which were labeled A and C.
Each animal in both of these groups was fed in the
individual feeding cages within between 5 and 10 minutes
after the individual training session.

Animals in group A were handled for a short period
by the experimenter and allowed to run free on a small

table exactly as in the habituation period. These

21

animals were fed 7 grams of food the first day and 4%
grams on the remaining days in the preliminary training
period.

Group C animals were placed in the starting box of
the experimental apparatus and allowed to run down the
maze straight alley to the pre-training goal box. The
animal was then allowed to eat for 30 seconds. The
experimenter attempted to time the animals only during
the period they were actually eating so as to compensate
for interruptions in eating in so much as this was
possible. If the animal did not eat within 5 minutes
after entering the goal box, he was removed and this
observation was recorded. This reward procedure was
followed in both the pre-training goal box and the
positive goal box for the remainder of the experiment.
These animals were given two massed trials each day.
They were fed 6% grams the first day and 4 grams on the
remaining days during the preliminary training period.

Consequently a very high drive level prevailed for all

animals.

E. Training

The preliminary training period was followed on the
next day by a four day training period. The training

groups were subdivided into three different training

23

training condition groups. These were labeled groups I,
II, and III which combined with the preliminary training
conditions gave a total of six primary groups-designated
I-A, II-A, III-A, I-C, II-C, and III-C.

Groups I-A, II-A, I-C, and II-C were equalized for
sex, initial position preference, and position of the
positive goal box. This produced eight possible
combinations in each of these primary groups. There
were 16 animals in each group giving two animals under
each possible combination of conditions.

Groups III-A and III-C were equalized for position
preference only but the other variables were equalized
as much as possible. There were 14 animals in each Of
these two groups.

Because of the equalization Of position preference
on the basis of the first training trial, most animals
were not assigned to a group until after the first
trial. This procedure made possible the important step
of beginning all groups at a level of exactly 50% correct
responses.

During the training period the animals were given
four trials per day for a four day period. Two of each
day's trials were to the negative end box and two were to
the positive goal box so that each animal had two

reinforced and two non-reinforced trials per day. The

23

trials of the first day were alternately free and forced
starting with a free trial. On succeeding days the
first two trials were free and the second two were forced.
The forced trials were arranged after the first day so
that only the following combinations of right and left
responses were possible: LLRR, RRLL, RLLR, and LRRL.
This was done to prevent an alternation pattern of
response.

The direction of each free response was recorded.
The animals were fed in the positive goal box in
exactly the same manner as in the preliminary training-
trials. On trials to the positive goal box the animals
were timed from the moment the first part of their body
touched the criterion line until they started eating.
They remained in the negative end box for 30 seconds
after entering it. On the first two days all trials
were separated by a period of approximately 30 minutes.
This time decreased somewhat on the second two days.

The animals were fed in the feeding cages between
5 and 10 minutes after their last trial for the day.
They were allowed to remain until they had finished the
day's ration. All animals were fed 4 grams Of food on
the first two days of this period and 5% grams on the
second two days.

Group I animals were trained consistently to one

side. For Group II animals the position of the positive

24

and negative end boxes was reversed after one free and
one forced trial only one of which was a correct or
reinforced response. For Group III animals the position
of the positive and negative goal boxes was reversed
after two free and two forced trials only two of which

were correct or reinforced responses.

F. Extinction

On the day immediately following the last day of
the learning trials all animals were given twelve massed
extinction trials. The only change made in the apparatus
for these extinction trials was to remove the food from
the floor Of the positive goal box. The animals were
allowed to remain in the end boxes approximately 10 sec.
for each extinction trial.

On all trials (preliminary training, training, and
extinction) the animal was forced to run after the
passage Of a reasonable period Of inactivity. This was
done by poking the animal with a long stick. This was
not recorded, if it was done in the starting box or
straight alley. However, in the few instances that
forcing was necessary at the choice point or immediately
in front of the choice point on a free trial, it was
recorded in the data. These few instances occurred

primarily during extinction and never on the critical

trials of the training period, i.e. trials 1, 2, or 3.
Animals were discarded from the data which refused
to eat within the alloted 5 minute period on the first
trial of the training period. They were also discarded
if they refused to eat on both rewarded trials of any
subsequent day in the period. Animals in preliminary
training grOup C were discarded from the data, if during
the preliminary training.period they refused to eat on
all trials of the first four days or if they refused to

eat on either Of the last day's trials.

V RESULTS AND DISCUSSION

The following measures were used for the
statistical analyses of the results: (1) number of
correct responses for all free trials for each animal
during training, (2) number of correct responses for
particular trials for all animals during training,

(3) number of correct responses on all extinction trials
for all animals, and (4) mean eating latency (time
interval between the instrumental response and the
eating response) per day per animal.

Analysis of variance of the total training period
demonstrates no significant differences between
preliminary training conditions or for the interaction
between preliminary training and training (see Table I).
Therefore, all further analyses of the training data
combine preliminary training conditions and inspect
differences only between training conditions. To assure
further that there were no significant differences
between preliminary training conditions a chi square
was calculated on the largest single trial difference
(Trial 2, Group II). This difference is not significant
as shown in item 1 of Table III. [Th3 training 9223,2212
not indicated any differences i2 learning 232.22
differential preliminary training conditions designed 32.
provide expectancies £23 gag g;ggp_and not for the gthgg.
Therefore, at this point hypothesis C is supported.

(26)

27

TABLE I

Analysis Of Variance Of Training Data for Preliminary
Training Groups and Training Groups in Terms of the
Number of Correct Responses per Animal*

 

Degree Off Mean Significant

 

Source of Variance Freedom. Square F P
Total 87 1.45

Preliminary Training 1 1.08

Training 2 15.55 11.29 .01

Position Preference X
Sex X Initial Placement

of End Boxes 42 .94
Preliminary Training X

Training 2 .15
Error 40 1.56

 

* These data meet the homogeneity requirements for
analysis of variance.

28

TABLE II

Group Comparisons Terms of the Mean Number of Correct
Responses for the Total Training Period

 

 

 

Correct Responses Actual Required
Groups N Mean 3. E. Diff. Diff. P*
I A&C 52 7.12
II A&C 52 6.25 .54 .87 .84 .02
I A&C 52 7.12
III A&C 28 5.65 .55 1.48 .97 .Ol
II A&C 52 6.25
III A&C 28 5.65 .55 .61 .6O .10

 

* P is based on a small sample t-ratio analysis.

29

TABLE III

Comparison of Preliminary Training Groups and Training
Groups in Terms of the Number of Correct Responses on
Pertinent Training Trials*

 

 

 

Copparisons Number Correct Chi

Groups Trial N Responses Square P
IIA 2 l6 5

IIC 2 l6 7 l.5l .50
Total 1 92 46

Total 2 92 6'7x 9.18 .01
I A&C 2 52 26

II A&C 2 52 10 14.27 .01
I A&C 5 52 28

III A&C 5 28 6 24.09 .01
II A&C 5 52 19

III A&C 5 28 6 7.59 .01

 

* A deduction of .5 from each discrepancy value has
been made to allow for the small frequencies and
emperical expected frequencies have been used in the
calculation of all chi squares.

x This figure represents the number of correct
responses based on the position Of the positive goal
box for the first trial. All other figures for the
number Of correct responses are based on the position
of the positive goal box for the indicated trial.

50

Analysis of variance of the total training period
given in Table I demonstrated a significant difference
among training conditions. Differences between the'
mean number of correct responses for the three training
groups are all in the hypothesized direction and the
significance of these differences is shown in Table II.
The probability of a correct response for all 92 animals
in terms of what was correct for the first trial
(position of the positive goal box on the first trial)
is well below the 1% level of confidence (see item 2,
Table III). The tendency to make an incorrect response'
on the second trial for Group II and on the third trial
for Group III, just after reversal of and box position
and in terms Of this reversed and box position, as
compared with Group I, which had no reversal, are both
well below the 1% level of confidence (see items 5 & 4,
Table III). The tendency to incorrect response for
Group III animals on the third trial just after reversal
of end box positions for that respective trial is
significantly greater than the tendency to incorrect
response for Group II animals on the third trial at well
below the 1% level of confidence (see item 5, Table III).

The same relationships during training are shown
graphically in Figure II. All the statistical analyses
together indicate rather conclusively that all training

groups differ significantly from each other in the

mamwha

 

 

 

m w b o m e m m H
OH
on
\I
x 11
use HHH .396 .III I \ II on
I- \
Uumq HH “509.0 .I'II \ ”>7
\ \ z 3
cam H macaw (I: \. \\\ . hJu
.\
mom I l I III! ITI IIIII. I on
\ V noncommem
1. \ . poshnoo
\
.\
.\\. Ob
\ \
\. om
\
\
, \ . , om
“s\-’II \ \ /\
\‘I \ 'l'k
.II‘ OOH

Hera. seem no momaoamom
pomuuoo Opeawpmb MO ammo new no magma a“ HHH use .HH .H museum mpfimOdsoo MO mm>aso mafiaawmq .N onsmah

52

hypothesized direction for the entire training period
and for the critical trials. In other words, reversal
after one or after two reinforcements retards learning
and all animals learned 9;; th_e_ M 9_f_ L135;___ first
reinforcement. At this point, therefore, expectancies
do not appear to be necessary to mediate learning and
hypotheses A and B are supported.

Analysis of variance of the total extinction data
does not indicate any significant differences between
the pertinent conditions nor is any significant
interaction between the pertinent conditions indicated ‘
(see Table IV).

The eating latency measure did not meet the
homogeneity requirements for analysis of variance when
based on mean time per animal. Therefore, the individual
within groups variance based on the individual variance
per day is used as the error term. This allows
comparison between preliminary training conditions and
between training conditions, and analysis of the
interaction between preliminary training and training
conditions as shown in Table V. A further analysis
of the differences between paired individual groups was
not possible with a total number of subjects this small
in view of the lack of homogeneity in the data.

Analysis of variance of the eating latency data

in Table V indicates that the only significant difference

55

TABLE IV

Analysis of Variance of Extinction Data for Preliminary
Training Groups and Training Groups in Terms of the
Number of Correct Responses per Animal*

 

 

Degree of Mean Significant

Source of Variance Freedom Square F

Total 87 6.24

Preliminary Training 1 2.78

Training 2 2.54 (There are no
significant F

Position Preference X values.)

Sex X Initial Placement

of End Boxes 42 5.87

Preliminary Training X

Training 2 1.89

Error 40 9 O 21

 

* These data meet the homogeneity requirements for
analysis of variance.

54

TABLE V

Analysis of Variance of Eating Latency Data for
Prelimdnary Training Conditions and Training Conditions
in Terms of the Mean Latency per Day per Animal

 

 

Degree Off Mean Significant
Source Of Variance Freedom Square F P
Individual 91 5,577.76
Preliminary Training 1 52,851.98 6.51 .02
Training 2 5,025.49

Preliminary Training X
Training 2 1,271.52

Individual within
Groups (error) 86 5,208.49

 

Figure 3.

55

Latency of Eating Response Curves fer

Composite Groups A and C in Terms Of the Mean Time

60

50

4o

30

Mean Time
in Seconds

20

10

per Trial per Day

 

Group A I, II, &III
-.........._ Group C I, II, 8: III
\
\
\
\
\
\
\
\
\
\
\
\
k“
‘\
\\
.\‘
x“
1 2 3 4

Days

36

results from preliminary training conditions. The graph
shown in Figure III illustrates this relationship. The
preliminary training conditions significantly affect

the time elapsing between the instrumental response and
the eating response. That isy those animals giypp
pgeliminary training gp.ppp_straight alley ppp
significantly sooner pppp_ppizp animals. However,‘pp$p
time difference appears 3p pp 13y pp affect learni g.
This lack of effect on learning is probably due to the
immediate secondary reinforcement resulting from the
Observation of familiar food, because the delay in
making the consummatory response is large for all animals
(see Figure III).

The apparent reason for the exactly opposite results
of no retardation of learning after reversal on the
second trial obtained by Moore is apparently due to the
lack of sufficiently immediate secondary reinforcement

after the instrumental response in her experiment. (18)

 

(18) Moore, pp. cit., p. 15.

 

A possible criticism.of the obtained results and
their interpretation could be drawn from an apparent
artifact in the experimental procedure. The apparatus
was designed in such a manner that the animal could
experience a stimulus pattern while he was eating that

would be the same as the stimulus pattern experienced

57

by the animal at the choice point before making the
instrumental response. The lack of difference on the
second trial in the amount of demonstrated learning
between trained (C) and naive (A) groups, or even the
increment on the second trial to the trained group

might be attributed to expectancies mediated by this
stimulus pattern. That is, on the second trial an
animal upon reaching the choice point might approach

the cues that have been associated with the eating of
food. This stimulus pattern would consist mainly of
floor cues and visual cues that would be different on
either side Of the choice point. HOwever, such an
explanation is a very remote possibility due to the

very close similarity of stimulus patterns on either
side of the choice point. Thgpp minimal stimulus
differences would pg pp Emil}. M ggneralization g:
reinforcement ppg inhibitionIQQEIQ completely nulify
any differential yplppflg£.ppp‘gpgp. Yet, the possibility
still remains that an animal responded correctly on the
basis of the close association of these cues with
reinforcement rather than on the basis of reinforcement
of the instrumental response. A sub-experiment designed
and performed to test this possibility follows. The
objective of the sub-experiment is to give just as great

an Opportunity for reinforcement of these minimal due

58

differences as in the main experiment without reinforcing

the instrumental response, and to discover if an

animal shows any tendency to respond appropriately on

the basis of these cues alone.

VII SUB-EXPERIMENT

A. Procedure

Exactly the same apparatus was used in this
experiment as was used in the main experiment. The
subjects were 12 male and 12 female albino rats from
the same rat colony. They varied in age from 80 to
105 days with.a mean age of approximately 95 days. The
habituation procedure was exactly the same as the
habituation and preliminary training procedure for
the naive (Group A) animals in the main experiment.
Exactly the same procedure of food ration was employed.
in order to maintain the same high level Of deprivation.

The training period was one day and it came
immediately after the habituation period. All animals
were placed in the starting box and allowed to run down
the maze stem to the choice point which.had the door
closed. Then the animals were lifted out Of the maze
stem by the experimenter and placed in either the positive
or negative end box and their respective maze arm.

Twelve of the animals, 6 of which were female and 6 male,
were each placed in the positive goal box and respective
maze arm with all doors closed. Three of each group of
males and females had the positive goal box on the right
and the remaining three of each group had it on the left.
All animals were then allowed to eat under the same

(59)

40

conditions as the animals in the main experiment.
Approximately one-half hour later each animal was run
down the maze stem to the closed choice point door and
was then placed by the experimenter in the negative end
box and allowed to remain there with all doors closed
for the same time interval as animals in the main
eXperiment. Exactly the same procedure was followed
with the remaining 12 animals except that they were
placed in the negative end box first and then in the
positive goal box.

Approximately one-half hour following the last
training trial each animal was given one test trial.
This test trial consisted Of allowing the animal to run
the maze with all doors Open. This was a free response
trial with the and boxes placed in the same relative
positions for each animal as they had been during the
training trials. The direction of this free response

was recorded.

B. Results and Discussion

The measure used was the number of responses in
the direction of the positive goal box for all 24 animals
on the test trial. The animals made 10 Of these
responses Out of 24 possible. The chi square of the

difference between this measure and the probability of

41

correct response on the second free training trial for
all 92 animals in the main experiment (67 correct
responses) is 45.29. This chi square with one degree
of freedom.is significant at far less than the 1% level
of confidence.

The animals in the sub-experimental group, in
other words, gave no evidence of correctly performing
the instrumental response on the test trial even though
the maze cues near the choice point were reinforced in
much the same way as they were for animals in the main
experiment on their first trial. Thereforg, y_e E‘l
conclude M 3133 learnipg that 3933 pl_._a_c_§_ gr; £12 _f_i_£§_t_
22121.12 the main experiment was due pg Qipgpp

reinforcement 32 the instrumental response and not 22

 

Egg approach‘ypgppflgg secondary reinforcipg‘gpgp;
Having removed this objection we may conclude that
all of the experimental hypotheses have been clearly
supported by the data. In addition, the data have
indicated at a high level of confidence that the
consummatory response (act of eating food) has no
effect on the learning of an instrumental response when
the Observed food has high secondary reinforcing value.
It then follows, that expectancies defined as
fractional-anticipatory-goal responses would not need

to be present in the organism for secondary reinforcement

42

to Operate. The naive animals learn on the first trial
as well as the animals with expectancy training. This

would be in conflict with the assumptions of Denny and

Moore (see pages 5 and 12).

It would appear that it is not necessary for the
instrumental response to have secondary reinforcing
value when all other forms of reinforcement are not
immediately contiguous with the making of the response.
This finding is exactly Opposite to the postulations of
-Spence and the experimental conclusions of Grice (see-
page 7). It would appear, therefore, that the primary
"gradient of reinforcement" originally postulated by
Hull has a very real existence. The "backward action"
of the reinforcing state of affairs on preceeding
responses and stimuli is presumably mediated by the
stimulus trace, and the stimulus trace is the basic

unit of the "goal gradient" just as Hull postulates.(l9)

 

(19) Po Be, p. 1580

 

The viewpoint that behavior is specified in terms
of a particular learning situation, solely by the action
of Hull‘s goal gradient, is effectively supported. It
follows, then, that particular need conditions have
far greater specificity than has been heretofore

supposed. This further suggests that the degree of

45

deprivation of a particular need may have a wide range
of influence in providing optimum conditions for
learning.

The mechanism of learning may be far less complex
than has been previously supposed. It is suggested that
all learning may be due to the temporal contiguity of
a reinforcing situation with particular stimulus traces
in the presence of a differentiated need condition,
uncomplicated by eXpectancies. This is essentially the
viewpoint presented in Hull's latest postulation

concerning primary reinforcement.(20)

 

(20) Hull, Clark L., Behavior Postulates and Corollaries
.2. November; 1949: p. 20

 

VII SUMMARY AND CONCLUSIONS

The subjects were 92 experimentally naive
albino rats. The apparatus was a modified T-maze
designed to make the time interval between instrumental
and eating response as short as possible. Half of
these animals were given preliminary training in the
maze stem in order to provide expectancies to the
experimental apparatus. The other half were given no
expectancy training. Each of these preliminary
training groups was subdivided into three training
groups. One training group had the relative position .
of the end boxes reversed after the first reinfOrcement,
the second had the end boxes reversed after the second
reinforcement, and the third group had no reversal.
After eight training trials all animals were extinguished
with 12 non-rewarded trials. A very high state of
food deprivation was maintained throughout the entire
experiment. .

A sub-experiment was performed with 24 similar
experimental animals to determine if the animals were
learning on any other basis besides direct reinforcement
of the instrumental response. No evidence of such a
learning mechanism was indicated.

The following results were found:

1. The learning Of both reversal groups was retarded

as compared with the group having no reversal, and the

(44)

45

group reversed after the second reinforcement showed a
greater retardation in learning than the group reversed
after the first reinforcement.

2. Preliminary training (expectancy training) gave no
better learning than no preliminary training.

5. Those animals having preliminary training ate
significantly sooner than naive animals but this time
difference does not seem to have any effect on learning
when almost immediate secondary reinforcement through
the perception of familiar food is available.

On the basis of these experimental findings we may
state that there is a very large increment to response
tendency after the first reinforcement in this
experimental situation, and that expectancies ppg‘pgp
necessary pp_mediate learning. Finall , pp mpy pp
postulated M all learning 32.123 p133 3p _t_h_e_ result
2; the temporal contiguity 9; pp; afferent neural
discharges resulting from.environmental stimuli pipp'p

neural discharge representipg reduction ip_p condition

 

‘33 deprivation.

46
BIBLIOGRAPHY

l. Grice, G. Robert, "The Relation Of Secondary
Reinforcement to Delayed Reward in Visual Discrimination
Learning", Journal of Empgrimental Psychology, vol. 38
(1948), p. I:I57""'

§§5ODenny, M. Ray, Unpublished lecture series (untitled),

3. Hull, Clark L., Behavior Postulates and Corollaries
2 November, 1949, p. I- 10.

4. Hull, Clark L., "Goal Attraction and Directing Ideas
as Habit Phenomenon", Psychological Review, vol. 38
(1931), Do 487'5060

5. Hullfi Clark L., "Mind,Mechanismi and Adaptive

 

gghavior , Psychological Review, vo 44 (1937), p. 1-
O

6. Hull, Clark L., Princi 1es of Behavior, New yorsz
Appleton-Century—Cro s, 0., I943. #22 pp.

7. Moore, Dorothy Nell, An Epperimental Stugy pf the

Effect pf Position ReversEI a er ne pp 0

Heififorcemefits on Sim Ie T-Maze Learnin IE the Rat,

Thesis for the pages 9: M; §;_(MIcHIgan Stfite COIIEge,
:PO 1" 0

8. S ence, K. W., ”The Role of Secondary Reinforcement
in De ayed Reward Learning", Psyphological Review, vol.
54, p. 1-8, (1947).

9. Watson, John B. The Wa s 9; Behaviorism, New York:
Harper and Bros., 19287- pp.

10. Wiener, Norbert, Cypernetics, New York: The
Technology Press, 1948. I94 pp.

 

 

APPENDIX

4s

zation layout for respective
lumn l is sex; 2 is position
is the position of the positive goal

box on the first training trial

groupS° ob
mm3 ’

e
,

Sex, position preference, and initial position of the

positive goal box equali

experimental
preference

II-C III-C

I-C

II-A III~A

I~A

.00

e

um 0

On 1234567891
a#

LRRRLLLLLLRRRR
RLLLLLLLRRLRRR
MMFMMFFMFFFFMM

LRRLRRLLRRLLLLRR
LLRLRLRRLRLRRLLR
MMFFMFFMMMMMFFFF

RLLRLLRRLRRLLRLR
RRLLLRRLLRLRLLRR
MMFFMFFMMMMMFFFF

RLLRRRRLLLRRLL
RRRLRLLLRLRLLR
FFPFMFMMMFFMMM

LLLRRRLRLRRRLLRL
RRLRLLRLLLLRRLRR
HFFMFFFFMMMMMMFF

RLRRLLLRRLRLLRLR
LRLLLLLRLRRLRRRR
FFMMMFFFFMMMMMFF

49

Summary of Animals Discarded from Data

Two animals refused to eat on all trials of the first
four days of preliminary training.

Three animals refused to eat on both trials of the last
day of preliminary training.

The following animals refused to eat on the first
rewarded training trial:

Preliminary Position Of Position
Training Gp. Positive Goal Box Preference

>00>00>
Hmwbfibw
wawaaw

 

The following animal refused to eat on both trials of
the third day in the training period:

Position of Position
Grogp_ Positive Goal Box Preference

IHA R L

 

 

.-
I
I
I
VI
I I.
I . .I‘
I \ 2",”
‘I l
I‘ ‘
I I V
I ‘ |
I ‘U
‘ . ‘1‘.
,“ ’3. ,I .
‘ ,II ' . I
;
I ‘s
‘ C
' I
b
l
I
I
I I
l
‘D 4 l
I' ‘ I .
‘0 I
I;-
l
".
1’ l ’
\
D ' I
I I 3’ . I-
I' I
I
I' I
I .
' I I 1 .‘
. . I
‘r‘ , ’ I .
I I
I I '
I ' \
. ' .I I
' I ,¢ . .
I ,2. ‘ ' O
I I l
g ‘I \ H . ' I
. I I . '
I ' '
- ‘ ' I
A - . ‘ , ‘ I
. ,, ., ,"’ I» I r .
III 1‘ .‘l . Ll ‘ l '
II ’ ' ‘ V I I
I I I ' ' I
‘ I“ ~ " m - v
‘3 I . ,
I ' ' I \ \ ’ --.
O J ' a s
l‘ ‘ :I ‘ l I Al " I
‘ I I ‘\" ‘ I '- . ‘,
IJ . .I . . N. V, . I - I \
w I ' . . -
l 4 r i I. ‘ ‘ .
, , I
.| p I» I .
I I 5 ‘I I A, , ‘ I
.r (I; s' It , I v I
i ._ ~ I .
I r-
" fir ‘ V, >1, " I ‘ I . ‘
v , , , ..
_IV‘I .. "‘ ' ‘j\.| I‘ I I
. .1" - :0 I
I l I" ‘i'.' ' .I ‘ " g I
R” ' II) 'I ‘ r. ' '
. ‘ I ‘
LI I I t I): M v‘ . ' .
. . ‘ ' i I I
I"! I‘ , \
fi’l é “ \I ‘2’ § .1“. \1 . I |
, I , y I .I .
I 1,} ‘I, “ l' 3, ‘ I I
I \ . " " -¢ ,
I ‘. ~ VI"! ‘ ." K I .' . I ‘
' ~‘. 3 ‘fv, :I‘)‘ ‘ . I- '
I j( . ‘ I
J I I ",)\ I 1, _ J ' . . , ,I
. 2 '. m I, n - . ‘ I .
. 1 I .. ,~ I . I K ) I
’ ,I ‘ ‘ 5 II I K I ‘ I , l
)4 'I‘ k). 84.. ’5 I 'l ‘ “ , It '
I ,1 I, II -J-~ . ’\ . .
A I ,1". ' Il..,"“‘" ' . _ , ‘
I v, . , M ‘ - , .
| I _ '..I I _; I I , . I
I I ' I . '“‘J"* ~I.". tr . ‘I H
I. r‘ ' ILWI' "“I.~“l'.\ 'L'r I .' ' ‘ II
_. , .I r.._.2 —1 I ' ' ' 0‘ a ' I
. J]. I‘. t: t. ‘\ I. I, .F ‘ _ I
..~.~;'.".',-.I\I'~~ II . ‘ ‘ ' ‘I ‘ “
.I ' I II' J ,‘I\ \ ‘ JI' 2) W ‘ \I 'l I ' _
' 4* I‘. . " ' '
h '1. ‘0 I- V" {I I |I I ' ‘ 1‘0 ' I - I . .
‘1‘ . s I t- v,‘ I . ‘
. ’1; 1‘... .\ . {.1 , I _ , ‘ . I . I. . - \ . ,
i I ' 4’4 ‘_ ', ‘ 1 I . I 1
AU ’4‘? I (:$ ‘1’. “ . ‘ I I I ' .
- . r, - I A" )I. ' ‘ ' l‘ ‘
_ 2 .v. 1. ‘L l, I ‘ w ‘
‘4' 'I‘ |'| . ,l- I j . I ‘ ‘ I ‘ I I
"III’I'Rn‘ "’ " 1.. " I‘ . ~ I I ‘ . . , I
' , ’I' I I" ‘ — ' ' I v r
I N J .I..{ I . ,. 1 ‘ I I I
I 'I f“: . ,I‘, ‘1 yd . I ‘ . .2
’ . . ‘2: y’0’:' ." ‘ ‘ l 'I | | ' I
. “ ' 1 I I ’
. I 1 I I “" .l I I \ b \ .
, III. I. . I 'I‘ .I O . I ' ..
I ‘ I I ’ , ' \. l‘ I, In I ' ‘ I I
‘ \ I I . r I” .1. g "I p . I J 'l - ' P_ I
.0 I, l ‘ v , '. .- \’ v I, l I . I. I I
. . p . l' ' ' v ‘I ‘I' II ‘ I
,I, I ". . ' M41" ’- , I
I I I I ' v , ( .4 v '
‘ I I ' I I ,
I I . .. ' I 7 3 l ‘ I,. I
I A I‘ . | I‘_ ..f' '1 ,‘ I .2 ‘1' , ‘W I "2 .‘ ' I
‘I \I I l ' ' . ' I .- ' ' ~
, I - o I ,7: I I $7 I I
. 'I .I, \ ‘ I ‘. . I I I I ‘ . -
I I I “I; -. ‘ L“ ‘I‘- I" '4‘ "1 ’CJ“), "1 I I, A . I A 1 .
‘ n , ', ‘ '.I ‘ ' , ., -
”WWW 3 LII-4' . ' I _ ' _ "mini-“'2 ’N‘“ ~ . 2., .. I2 . ‘I.

-1- ._._ __“

 

_—_"

 

 

J
.l
I v
.
. I . I
. \ 9
N. I
l I I J.P I u
I . J I I .‘ . 5
4 I I l o O o l .a c
. .’I .l . u . .
It i II
(. II . \ . I
I,
I I. . ,Ibll \ . I I . a . I.
J .. ‘I. Q” Ian I III. ' I v."..‘I‘AOII I To nl
. vulv- .\I. aII \I. .. p. Ora... Inns J .
. .v... (final Ifi... I Mfr .nv . . .I1.I ..wMU.J
. I I .

F. (...! I I
II JI“..1¢1’..1..?.... A“...hfl»\......

   

7m GAN ISTATE UNIVERSITY LIBRAR

as
61, I

I

I