_E_:____ __S_“_;__;_T: 3::

 

. Yx

. 1s
7 a
F ¢
. ...
o . u
‘ ..
e u
~O .-
o
gm.
. g
«4...
A
a;

A GESTALT VERSUS A STIMULUS
RESPONSE ANALYSIS OF REASONING
IN THE RAT

B!
Anthony M. Graziano

A Thesis
submitted in partial rulfnllment of the
requirements for the degree of
Master of Arts
in
Michigan State University

Department of Psychology

1956

5>A2% 537

7’ #330

The author wishes to express his sincere
thanks to Dr. M. Ray Denny, under whose direction
the present investigation was undertaken. His
assistance and guidance throughout the eXperiment
and thesis preparation have been deeply appreciated.

II.
III.
IV.

VI.
VII.
VIII.

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . .
LIST OF FIGURES . . . . . . .
HISTORY . . . . . . . . . . .
INTRODUCTION TO THE PROBLEM .
STATEMENT OF THE PROBLEM . . .
METHOD . . . . . . . . . . . .
Apparatus . . . . . . . . .
Subjects . . . . . . . . . .
Procedure . . . . . . . . .

Handling 0 e s 0 e o e o 0

Acquisition series, foodawater

AOQuisition series, water-food

Test trials . . . . . . .
RESULTS . . . . . . . . . . .
AEIperimsntal Group I . . . .
Experimental Group II . . .
Control Group . . . . . . .
DISCUSSION . . . . . . . . . .
SUMMARY’KND'CONCLUSIONS . . .
BIBLIOGRAPHY . . . . . . . . .

sequence

seQuence

PAGE

11

14
16
16
2O
20
2O
20
22
23
24
24
24
26
29
37
39

LIST OF TABLES
TABLE PAGE

I. FREQUENCY OF RESPONSE AND CHI SQUARE
VALUES AND PROBABILITIES FOR.AEL
GROUPS ON EACH TEST TRIAL . . . . . . . . . . 25

II. COMPARISON OF GROUP II AND CONTROL
GROUP ON EACH OF THE THREE TEST
TRIMS O O 0' O O O O O O O O O O O O O O O C 28

LIST OF FIGURES

FIGURE PAGE

1.

2.

3.

4.

One or the mazes used in maier's “reasoning"

A.

eXperimont. Animals were allowed to ex- -

plcre the mass and to approach food at P,

without the barriers of the wall and wire

cage. Then with the cage and wall in

position, the animal was placed at A. Oh-

able to go directly from A tolF, the. animals

showed I"'reasoning by scaling the wall,

entering box b, and taking the correct path

to the 3031,}? . Meier, Comparative? Pszcholggz
Monographs, 1929, 29:11. . . . . . . . . . . . . . 8

conventionalized form of a mass used in
Meier's "reasoning" eXperiments. Hull,
Psychological Review, 1935, 42: 222. . . . .p. . . 11

The mass used in this study. Animals were

trained to run.Rdx and Uéx for food reward,

and then R-U and R-H for water reward.

Another group or animals was trained in the

same manner but with the water trials first.

On the test trials for all groups, the

animals were placed hungry at R, with alleys

I and III open. Entering alley I constituted

a correct choice, and alley III an.incorrect -
6110105. 0 O O O O O O O O 0 O . O O O O O O O O O O 17

The mean daily response time for each group

on both water and food acquisition trials . . . . 54

Ill-Ill

I. HISTORY

The psychology of Animal learning . . . has been and
still is primarily a matter of agreeing or disagreeing
with Throndike, or trying in minor ways to improve upon
him. Gestalt psychologists, conditioned-~reflex psycholo-
gists, sign-~gestalt psychologists—-all of us here in
America seem to have taken Thorndike, overtly or covertly,
as our starting point. (1, pg. 19)

At the end of the last century, on the basis of the
puzzle-box experiments and others, Thorndike established a
theoretical position of trial and error learning. He sum-

marized his eXperiments in Animal Intelligence (7), and con~

 

tended that in the random movements of the cat in the puzzle-
box the correct response is eventually made by chance alone.
Then, through a trial and error process of "stamping in“ and
"stamping out,“ the correct reaponse is strengthened, and the
incorrect responses are weakened.

This strengthening and weakening was, in part, referred
to by Thorndike as the law of exercise. This includes the law
of use, Which is the strengthening of connections with prac-
tice; and the law of disuse, which refers to the weakening of
connections when practice is discontinued.

Another of Thorndike's major formulations is the law
of effect, which states that the consequences of a response
have an effect upon the strengthening or weakening of that
reaponse. The strength of the response is increased when
that reaponse is followed by a satisfying state of affairs.
Hilgard (l, p. 24) quotes Thorndike as writing:

2

By a satisfying state of affairs is meant one which
the animal does nothing to avoid, often doing things
which maintain or renew it. By an.annoying state of
affairs is meant one which the animal does nothing to
preserve, often doing things which put an end to it.

Thus, the correct response occurs by chance and is
strengthened through repetition and the satisfying state of
affairs following it. The incorrect responses are weakened
because of the annoying state of affairs which follow them
and are further weakened by disuse. Hilgard (l, p. 47)
wrote:

The flavor of Thorndike's theory has all along been
that of the automatic strengthening of specific con-
nections--direct1y without intervening ideas or conscious
influences.

At least one constructive result, indirectly, came
out of world war I, when.Kbhler was confined on the British
controlled island of Tenerife. There, with his assistants
and a cooperative colony of chimpanzees, he studied the
problem solving behavior of chimps, and applied gestalt
theory to the field of learning.

Kohler (4) described the intelligent problem solving
behavior of his animals, and criticised Thorndike's trial
and error position, and his conclusion that animals are not
capable of reasoning or insight. Kbhler felt that since
cats are not ordinarily confined to boxes and required to
push buttons in order to escape, the entire problem situa-
tion utilized by Thorndike was an unnatural one. The cor-

rect response in this situation, therefore, could be attained

in no other way but by chance, since the response was not
'within the animal's usual repertoire of responses.

The puSZIOJbOI situation.was also one in.whioh the
animal could not see all of the necessary components for the
solution of the problem. Kchler (4, p. 3) wrote:

As experience shows, we do not speak of behavior as
being intelligent when human beings or animals attain
their objective by a direct unguestionable route which
clearly arises naturally out of their organisation. But
we tend to speak of "intelligence“ when, circumstances
having blocked the obvious course, the humanfibeing or
animal takes a round-about path, so meeting the situation.

He then conducted experiments with chimpanzees in
which the "direct way" was blocked and the animal had to
utilise a around-about way" in order to overcome the block
and attain‘the objective. “Kohler also required that the
objective, the obstruction, and .13. the total field or
round-about routes be in plain sight, and the required acts
be within the normal repertoire of the animal.

One of Kohler's most quoted experiments, and one which
illustrates the occurrence of insight and intelligent be-
havior by a chimpanzee, is the double-stick problem; which
has, incidentally, also immortalized Sultan. Kohler (4, p. 125)
described this situation thus:

This time Sultan is the subject . . . his sticks are
two hollow, but firm, bamboo rods . . . the one is so»mmch
smller than the other, that it can be pushed in at either
end of the other quite easily. Beyond the bars lies the

'obJective, Just so far away that the animal cannot reach
it with either rod. Nevertheless he takes great pains to

try to reach it with one stick or the other, even pushing
his right shoulder through the bars.

4
Sultan continued to try to reach the fruit, making bed
errors, and some ”good errors;“ (4, p. 125) such as when he:

. . pushes one of the sticks out as far as it will go,
then takes the second, and with it pokes the first one cau-
tiously towards the objective, pushing it carefully from the
nearer end and thus slowly urging it toward the fruit.

‘ . . thus, all of a sudden, for the first time, the contact
animalnobjective'l has been established, and Sultan visibly
feels (we humans can sympathise) a certain satisfaction in
having so much power over the fruit that he can touch and
slightly move it by pushing the stick.

After watching Sultan's failures for over an hour,
Kohler understandably left the task of observing to a keeper,
who reported: (4, p. 126)

Sultan first of all squats indifferently on the box

. . . then he gets up, picks up the We sticks, sits down
again on the box and plays carelessly with them. while
doing this it happens that he finds himself holding one
rod in either hand in such a way that they lie in a
straight line, he pushes the thinner one in a little way
into the Opening of the thicker, jumps up and is already
on the run towards the railings, to which he has up to
now half turned his back, and began to draw the banana
towards him with the double stick.

The conditions under which Sultan solved his problem
illustrate insight and the use of a round-about path to an
objective. In this instance, the direct path to the objective
(fruit) was blocked by cage bars, and an indirect path of
first putting two sticks together, and then combining this be-
havior with that of reaching for the fruit with the longer
double stick had to be taken.

In Kchler's experiments, although he does not explicitly

set up a theoretical framework complete with postulates, cer-

A‘flln'llllll'lll‘“!

tain.consistent underlying notions become evident. Iohler
(4, p. 11) wrote:

When.any of these higher animals, which make use of
vision, notice food (or any other objective) somewhere in
their field of vision they tend--so long as no complica-
tions arise--to go after it in a straight line.

when this direct path is blocked, there is a tension

created in the organismfis psychological field, and this
tension varies with the need for the goal and its distinc-
tiveness from the background. unresolved tension creates
conductive forces, tempered by the physical barriers, or re-
straining forces of the geographical field. Thus Sultan per-
sisted for over an hour, hovering near the cage bars with his
two sticks, in.futile attempts to reach the fruit.

In his discussion of the gestalt interpretation of

problem solving, Osgood (6, p. 607) wrote:

The unstable, fluctuating pattern of tensions in the
psychological field tends to be minimised through sudden
reorganisations, involving the perceptions of new paths
to the goal.

This sudden reorganization of forces in the psycho-
logical field is the occurrence of insight.

In describing the characteristics of insight, Osgood
listed suddennesa as the most obvious characteristic of an
insightful solution. This differs from trial and error
solutions, or usual conditioning, which are ". . . a gradual
and irregular accretion." ‘

Smoothness is another characteristic, and was illus-

trated by Sultan's quick, smooth transition from sitting on

 

[It‘ll‘lu‘ll'l'lll'lll‘

6
the box to being '. . . already'on the run toward the railing."
Once the solution was apparent to the animal, there was no
longer ambivalence or bad errors,‘but a smooth flow of behavior
‘which resulted in success.

In the occurrence of insight, the solution preceded the
execution of behavior: as contrasted with Thorndike's cats,
which first, by chance, released the door-opening mechanism
before learning the solution.

This seemed to be the most important characteristic of
insightful behavior, as demonstrated by Kohler's animals, and
was illustrated many times over with dogs, as well as with
chimpanzees.

Here lies the primary difference between the approaches
"or Thorndike and Kohler. In the first, the animal's random
movements bring about success, and are gradually narrowed
through exercise and effect to the most effective behavior.
The solution is the.meohanioal association, which is strenth-
sued by the succeeding satisfying state of affairs.

In Kchler's formulation, however, even though he des-
cribes considerable trial and error behavior, this behavior
does not seem to be completely random, and the recognition
and solution of the problem precedes the experience of success.
Arriving at the solution, then, does not depend upon what
follows the behavior; but rather the solution precedes, in

temporal secuence, the experience of success.

II. INTRODUCTION TO THE PROBLEM

In 1929, Meier (5) reported his work on the problem-
coivirg behavior of white rats, from which he concluded that
the 'round-about-way' problems involved the formation of
Gestalten. Meier constructed situations in which the animal
could become familiar with a variety of paths; some of which
led to the goal,_and some of which did not. Then the primary
path to the goal was blocked, and the problem for the animal
was to utilize these experiences and arrive at the goal.

One of these mazes is shown.in.Figure 1, page 8. In
this experiment, the animal was allowed to become familiar
with the table top, and given practice in scaling the wall
and exploring the various alleys, including the correct path.
The animal then learned to run.correct1y from a box at point
b, to the goal, F; which was inaccessible by any other means.
The wire cage extended around it, but was open at the juncture
of the alley and F. In the test situation, the animal was
placed next to the wire cage at point A. Meier reported that
after unsuccessfully attempting to get through, over, or
around the wire cage, the animals all eventually solved the
problem by leaving the area around A, scaling the wall, going
directly to box b, and then taking the correct path to F.

After a number of experiments similar to this, flaier
concluded that the solution of the problem illustrated the

formation of patterns or Gestalten. In the experiment here

 

 

 

 

S

fl

«1
'VVflkLl.

 

 

 

 

 

 

 

 

 

 

Figure 1. One of the mazes used in maier's "reasoning"
experiment. Animals were allowed to explore the maze and to
approach food at F, without the barriers of the wall and wire
cage. Then'with the cage and wall in position the animal was
placed at A. unable to go directly from A toiF, the animals
showed Hreasoning by sealing the wall, entering be: h, and
taking the correct path to the goal, F. Meier, Comparative

Paycholgfiz longgraphs, 1929, 29: ll.

k! lllll‘lvl‘ullll'll'llllvll“

9
described, for example, the solution did not depend upon trial
and error since the solution seemed already to have been ar-
rived at when the animal left point A, scaled the wall and
entered box b. Also, if we were to base an explanation on
simple association, we would have to assume that point A is
equally well associated with all other points on the maze.

An explanation based on pure association, then, cannot account
for the fact that the animals correctly chose the a to b run,
rather than A to any other point.

Meier explained his results on the basis of the animals
having experienced the following patterns: (1) A to F; (2) the
pattern of the maze in general from exploring it; and (3) b to
F. To solve the problem, a new path, Aab-F, must be formed
through combining parts of the other patterns. This is a new
combination of separate experiences and is not formed merely
by repetition. Insight is the formation of this pattern,
which is the solution, and does not depend upon trial and
error, or simple association. Meier (5, p. 92) wrote:

The concept of patterns or Gestalten thus seems to be

a necessary assumption to explain these complex types of
behavior. The fact that a rat can choose the shorter
means to an end without previously having reached this
end by any of these means, seems to make a pattern con-
cept almost a necessity. A temporal chain is not eat.
ficient, it must be an immediate whole.

Hull agreed that the solution of such problems depends
upon more than simple association, or trial and error. He

wrote: (2, p. 220, 21)

hut...

 

10
. . . the genuinely creative and novelty-producing
portions of the reasoning process take place in advance
of the emergence of the substance of materials of the
Syllogism; 1.9., the solution consists in the assembl ,
from the considerable store of such materials presumeEly
possessed by the more versatile and adaptive organisms,

of the particular set of premises which are relevent to
the problem situation in question.

Chance is evidently an element in intelligent behavior
but, clearly, the dice of chance must be loaded in some
way or problems would never get solved. In short, any
adequate theory of higher adaptive behavior must show how
the dice are loaded; 1.6., how the characteristics of the
problem situation are able to evoke the particular combin-
ation of acts which alone will serve to extricate the or-
ganism from its difficulty.

However, Hull felt that, although Meier had prOperly
rejected the simple association eXplanation, his utilization
of Gestalten was not explanatory. Meier had not demonstrated
how these patterns were formed to bring about the solution.
The real problem was to deduce the formation of the solution
by the combination of behavior segments from more basic
principleS.

Hull then set up a conventionalized form of Meier's
"reaSoning" eXperiments; Figure 2, page 11. R is the starting
box; while'U, x, and H are goal boxes, each with its own
distinctive stimuli. Animals would be taught, through forced
tria18, to run from R to X, and from U’to X for food reward.
After this is learned, they are taught to run, food satiated,
but thirsty, from R to U, and R to H for water reward. In

the test trials the animals are placed hungry at R, with the

 

 

.Illllal. ’wllrl'llllllllll'l‘ll’lll ll '1‘! III

11

 

 

 

Figure2 . A conventionalixed form of a maze and in
[nier's "reasoning" experiments. Hull, PeIchological Review,
1935, 42: 222. ""'"""

12
previously-learned direct path, R-x, blocked at point B. The
problem is for the animal to reach 1, and the solution, of
course, is to run the sequence R-U-x. That the animals do
solve this type of problem is illustrated by Maier's experi-
meats. I .

The essence of the problem as Hull views it is to show
that the stimulus complex of R is more closely associated
with the R-U response than with the ma response. In learning
the U to x path for food reward, the stimulus complex at U
comes to elicit anticipatory food-taking responses, which are
strengthened with each repetition of the UAX run. Thus, each
tine the animal is placed in the U goal box, it will respond
to these distinctive stimuli by making the anticipatory goal
responses and then the "correct“ response of running the U
to X path. Then, when running under water deprivation along
the R-U path, the anticipatory food-taking responses, elicited
in the past by the stimuli of the U goal box, are again elic-
ited when the animal enters this goal he: and are strengthened
by the abatement of the water drive. Through repetition of
reinforced R-U trials, the anticipatory food-taking responses
become associated with allay I, and eventually with the R-U
choice point in the starting box, R.

In contrast,with this, goal be: H has never been
associated with food. Anticipatory food-taking responses will
then not be elicited by the stimulus complex of goal box H and,
of course, will not become associated either with alley III,

13
or with the R-H choice point in the starting box. Then, when
the animal is placed hungry in the starting box, the pro-
prioceptive cues contingent upon food deprivation will elicit
anticipatory food-taking responses; and since these are al-
ready differentially associated with the R-U choice point,
the animal will tend to make this response rather than the
3-H response. Once in goal box U, the problem is solved and
the animal Quickly. runs the previously learned U-X path and
arrives at the goal.

In this way Hull explained the solution of the problem
by the animals without referring to any higher mental proc-
esses as reasoning or insight, but by deducing the Operation

of intermediary S-R mechanisms as representational processes.

III. STATEMENT OF THE PROBLEM

Osgood (6), in his discussion of Hull's treatment of
the problem, points out that this explanation depends upon a
certain sequence of training. That is, the anticipatory food-
taking responses must be already associated with.the stimuli
of box U in order for them to be reinforced on the water trials
to U and then to become associated with alley I (see Figure 3,
page 17) and then chained back to the RPU choice in the start-
ing box. If the training sequence were reversed, if the water
acquision trials preceded the food acquisition trials, then
the fractional anticipatory goal responses elicited by the
stimuli at box U would not become associated with alley I, and
the RrU choice in the starting box. There would be, then, no
loading in favor of the RrU choice, and in the test situation
the animals should.make the correct choice with no better than
chance success.

According to Maier's explanation, however, the sequence
should have no effect upon the test choice, since the formation
of gestalten does not depend upon the sequence, but upon the
combination of parts of separate experiences into a new and
correct,pattern.

Utilizing Hull's conventionalized form of Maier's
reasoning experiment, the experiment described by Osgood and

reported above, will be performed. This experiment is so

15
designed as to test the explanations of both Hull and Meier.
All groups, according to Meier's explanation, should solve
the problem through the formation of gestalten, while Hull
would predict that only the group trained in the food-water
sequence will be able to solve it. Thus, if neither of the
two groups, or, if only one of the two groups makes the correct
choice with significantly better than chance success, than we
would have to reject Meier's gestalt interpretation and conclude
that there is no reasoning or insight evident in the animals'
behavior. .

Results supporting Hull's view would be that the group
'trained in the food—water sequence makes the correct choice
with significantly better than chance success, while the other
groups do not. Any departure from these results would lead us
to reject the efficacy of the fractional anticipatory goal re-
sponses in bringing about the solution.

If, of course, none of the groups solve the problem,
the conclusion will be that neither the formation of gestalten
nor the mediation of the fractional anticipatory goal responses

were operative.

IV. METHOD
Apparatus

The apparatus used in this study was a four-alley
wooden.mase as shown in.Figure 3, page 17. The mass was con-
structed as a single stationary unit with no interchangeable
parts, and included a starting box (R), with Openings into
three straight alleys (I, II, and III); three distinctive
goal boxes (U, x, and H); and a straight alley (IV), which
connected the U’andlx goal boxes. a small platform.(i), with
enclosed sides, was constructed which led into the starting
box so as to insure the animalls entering the starting box .
with essentially the same position orientation on each trial.
The dimensions of the maze are as follows;

R = 1:5 in. wide by v in. long by 12 in. high

A.8 5 in. wide by 8 in. long by 6 in. high

I z s in. wide by so in. long by 12 in. high
II 8 5 in. wide by 23 in. long by 12 in. high
III = 5 in. wide by so in. long by 12 in. high
IV = s in. wide by 14 in. long by 12 in. high

0 a s in. wide by 12 in. long by 10 in. high

X.= 7 in. wide by 13 in. long by 12 in. high

H = 6 in. wide at the entrance and 12 in. wide

at the rear by 8 in. long by 12 in. high

7
1.

. . .oodono essence
is.“ so HHH hogs use 636% possess a neuapauesoo H hens 93.395 5er HHH
one H shone and... .m an hnwssn cocoa each massage one .25on Has no.“ 33.5 one»
on» so .35.“ 33.3 none: on» near use posses case on» a.“ @2395 new edges
no macaw suspend 6.3.39“ nope: you. Mum one Dam s25 use .cnsaen coon Hon ”up

can gum 5a on condone ones 3334 .25: nan» a.“ coma ones ens an 0.33m

 

     

«can Ill:

'-"

233.4730 .l\/\l\

 

 

 

 

 

 

 

D-ELO D

W e urn}
QfiLca

 

18

The reason for the high sides on the maze was to slim-
inste extra-maze cues and the learning of simple approach re-
sponses by the animals.

The sides and the bottom of the maze were constructed
of three-fourth inch plywood. The top of the maze was
covered with one-fourth inch hardward cloth hinged along one
side to facilitate easy access to the running animal in case
the necessity should arise. Guillotine doors were located
as shown in.Figure 3, page 17, at the entrance of each dif-
ferent section of the maze. A system of strings and counter-
weights was connected to the doors and attached to overhead
slats So that opening and closing the doors could be accom-
plished from one central position.

The entrance platform and starting box were of untinted
natural wood sanded smooth and the sides and the bottoms of
the straight alleys were painted a uniform flat grey. The
entrance platform extended two inches into the starting box
and a wire fence was placed as shown in Figure 3, page 17, to
prevent the animal's doubling back past the end of the starting
platform‘within the starting box.

Each curtain was a single piece of grey cloth of the
same shade as the sides of the alleys and was placed in each
alley eight inches from the starting box. in attempt was made
to have the portions of the alleys between the starting box

and the curtains as homogeneous and as similar to each other

19
as possible to prevent the interference of extraneous cues
which may have affected the animal's behavior.

An essential condition of this Study was to maximize
the distinctiveness of the goal boxes so as to provide for a
clear association of the cues of the stimulus complex of each
goal box with the particular appropriate reward. To provide
this condition, goal box U was rectangular in shape, painted
white and had a floor of soft, perforated fiberaboard material.
Goal box x was also rectangular in shape, but larger than U,
was painted the same flat grey as the running alleys, and had
a hard, wooden floor. Goal be: H was trapezoidal in shape,
being narrower by seven inches at the entrance than at the
rear, was painted black and had a sheet of tin covering the
floor.

In goal box x a metal dish was attached to the floor
to prevent the animals from dragging the dish out of the goal
box. Holes were drilled into and wire supports were attached
to the sides of goal boxes U and H to provide for the inser-
tion of the water bottles and drinking tubes. The exact po-
sition of these is shown.in Figure 3, page 17.

The maze was securely attached to the top of a table
of such a height as to facilitate easy access and observation
by the experimenter. Constant illumination was provided
throughout the experiment by fixed overhead lights.

20

Subjects

The subjects for this experiment consisted of 37
experimentally naive albino and hooded male and female rats
from the colony maintained by the Department of Psychology
of Michigan State University, East Lansing, Michigan. The
ages of the subjects at the beginning of the experiment
ranged from 82 to 208 days, with a mean age of 109 days.

[procedurg
The procedure falls into four main categories: (A)
handling; (B) acquisition series, food-water sequence; (0)

acquisition series, water—food sequence; and (D) test trials.

Handligg. Seven days before the initial trials, each
group was handled by the experimenter. They were picked up
and stroked and generally manipulated so as to reduce the
emotionality of subjects and experimenter to insure the
successful functioning of both during the eXperiment.

During this seven-day period, the animals became
relatively docile and the marked aversive responses to the

experimenter diminished considerably.

Agguisition series, food-water sequence. At the be-
ginning ef the seven-day handling period, experimental Group I
was given fortyaeight hours of food deprivation. For the re-
maining five days each of the ten animals received eight grams

of Purina Dog Chow, mixed with.an equal volume of water at the

21
hour of the day in which it was to be run in the mass. This
procedure maintained a controlled hunger rhythm of about
twenty-four hours.

The acquisition series for experinwntal Group I ex-
tended over an eighteen day period. The subjects were given
four trials apiece each day for seven days, with no trials on
the fourth day. The trials were spaced so that all the sub-
Jects were run before the first animal was run a second time.
The animals were water satiated during the acquisition trials
in which food served as reward.

On the first day, the animals were given four forced,
spaced trials from R to goal box x with food reward. The
doors, located as shown in Figure 3, page 17, were closed as
soon as the animal had passed them to prevent retracing. The
animals were allowed thirty seconds of eating time in the goal
box. The food reward consisted of ground Purina Dog Chow,
mixed with an equal volume of water.

On the second day of the acquisition series, the animals
were given four spaced trials apiece on the U to X alloy, with
the food reward and the eating time the same as on the previous
day.

On the succeeding five days, the animals were given We
trials from R to I and he trials from U to. X. All trials
were spaced and the reward and feeding time remained the same.
Bah animal ran these alloys in alternate order.

22

After seven days of this learning, the animals were
fed to satiation, put on twenty-four hour water deprivation,
and then were run on the 3-H and R-H alleys for water reward.
After ten days of this training, the animals were water
satiated, put on twenty-four hour food deprivation and given
We more U-X trials for food reward.

This group began to show marked aversive behavior to
goal be: H on the water reward trials. This persisted over
a period of four days and reached a point where the animals
would not enter alley III which led to goal be: E. This sit,-
uation was corrected when the animals were well satiated with
food before making the water runs.

The reason for this differential behavior to goal be:
H was not fully determined, but it was felt that it might
have an effect upon the subsequent test trials. For this
reason, we added a control group which was treated exactly
the same way as experimental Group I, except that all animals

were well satiated for food before making the water runs.

Acquisition 33M, water-food sequence. Experimen-
tal Group II, which consisted of thirteen aninnls, was
trained first to run for water reward, and then trained to
run for food reward.

They were food-satiated, but thirsty and were given
two forced, spaced trials on R-U and two on R-H, each day

23
for ten days. They were rewarded with sixty seconds of
drinking in each goal box.

After ten days of this training, they were allowed to
become water satiated and were then put on forty-eight hour
food deprivation. The acquisition trials for food reward for
this group are the same as those described for experimental

Group I.

Test trials. Bch group was given a test trial twenty-

 

four hours after the last food trial of the learning series.
For each group, the test trials consisted of having both 3-H
and R-H alleys open with R-x blocked. fish animl was placed
hungry at the entrance alley a and was faced with a choice
between 3-0 and 11-3. after all animals in a group had been
given one test trial, they were given another. On the next
day, each animal was again tested once. The animls were
allowed fifteen seconds of feeding time upon reaching goal
box I, to avoid extinguishing the response. They were also
allowed to retrace from the incorrect H goal box to the
correct goal box, X. Thus, each animal was fed for fifteen

seconds in the x box on each test trial.

v. RESULTS.

uperinmntal 931932 I

is can be seen from Table I, nine of the ten animals
in this group made the correct choice on each of the three.
test trials. on the first test trial, .11 of the animals
approached both of the open alleys, and all but one showed
marked recoil from the R-H alley. Progress along the R-U
alley was rapid, the only pause occuring once in the U goal
box. There, each of the animals which had made the correct
choice approached the end of the goal box where the water
tube had been previously, investigated the hole in the side
of the box and then turned and went to goal box I. The be-
havior of the aninals in test trials We and three was
essentially the same as in the first, but with a decrease in
the time required to reach goal box 1:.

Utilising a chi square design with expected fre-
quencies of five correct, and five incorrect choices, it was
found that chi square, for one degree of freedom, equalled
4.90, with a probability less than .05 but greater than .02.
Thus, in each of the three test trials, Group I made the

correct choice with significantly better than chance success.

Experimental Group I;
This group, trained on the water-food sequence, did
not make the correct choice with significantly better than

TABLE I

FREQUENCY OF RESPONSE AND CHI SQUARE
VALUES AND PROBABILITIES FOR
ALL GROUPS ON EACH TEST TRIAL

 

 

 

 

 

A}; 2”“; :
Trial Group I Group II Control
Number Group
One R w R. W R W
o 9 1 0 8 7 0 7 5
E 5 5 E 7.5 7.5 E 6 6
x2 = 4.90 x2 = o x2 =- .042
d. f. = 1 P = .50 d. f. = l
.05) >.02 .45>P>.1+0
Two R I R. W R. I
0 9 l O 10 5 0 9 3
E 5 5 E 7.5 7.5 E 6 6
x2 - n.90 x2 .. 1.066 x2 = 2.08
d. f. a 1 d. f. a 1 d. f. = 1
Three R w R W R W
O 9 1 0 11 h 0 10 2
E 5 5 E 7.5 7.5 E 6 6
:2 - 14.90 x2 = 2&0 x2 = 4.08
o. f. = 1 d. f. ='l d. f. = l
.05>P>.02 .10) >.05 05>P>.02

 

 

 

 

 

right (to U)
wrong (to H)

observed frequencies
expected frequencies

HO 1:313

26
chance success on any of the test trials. 0n the first test
trial eight of the fifteen animals made the correct choice,
ten of the fifteen on the second, and eleven of the fifteen
on the third. From the probabilities given in Table I, page
25, it can be seen that none of these results are significant.

The behavior of this group was different from that of
Group I in that they did not show the marked recoil from the
R-H alley as did the first group. Whereas most of the animals
in the first group approached the R-H alley, recoiled from it
and then went directly to the R-U entrance, those of Group II
either entered the starting box, paused about midway between
the two alleys and then went directly to one of them, or they
came off the entrance alley and without any pauses went di-

rectly to one or the other of the alleys.

Control‘ggggp

The Control Group, which was trained on the foodswater
sequence as was Group II, did not make the correct choice with
significantly better than chance success on test trials one
and two, with seven out of twelve, and nine out of twelve
correct choices respectively. From Table I, page 25, it can
be seen that for trial one, a chi square of zero gives a
probability of .50, and for trial two, the probability is
between .25 and.15. On the third test trial, however, ten

of the twelve made the correct choice. Chi square equals

27

4.08, and with one degree of freedom the probability lies
between .05. and .02.

The behavior of this group was essentially the same
as that of Group II, again differing from Group I in the
absence of the recoil behavior from the R-H alley.

Since the correct choices of trial three of the control
Group showed significance, the question was asked whether this
group made the correct choice significantly more than did
Group II. The two groups were compared directly on each of
the three test trials, and as seen from Table II, page 28, '

there are no significant differences between the two groups.

28
TABLEIII

commsor OF GROUP- II AND CONTROL GROUP
0s FACE OF THE THREE TEST TRIALS

 

 

 

Trial Trial Trial
One Two Three
R H R a B II
Group Group , Group
II 8 7 II 10 5 II 11 4
Control Control Control -
Group 7 5 Group 9 5 Group 10 2
X2 C .017 do to . 1 do fo . 1
do to . 1 p . o“21* p - o4841*
.45)P>.40

W

Eisner 's exact text .

 

'III'. II

VI. DISCUSSION

The results of this study, as reported in the previous
section, do not support either Meier's or Hull's position.
According to Meier's analysis, all groups used in this study
should solve the problem through the formation of gestalten.
That is, the emergence of new I'wholes" in an insightful
solution depends upon the experiencing of all the necessary
sections of the maze and not upon the order of these ex-
periences. Thus, all groups used in this study should have
solved the problem if the formation of gestalten was truly
operative, since the animals did traverse all the alleys.
That Group II and the Control Group did not solve the problem
is evidence against Meier's gestalt position. There was no
evidence of insightful solution and maier's analysis in terms
of the formation of gestalten in such problem-solving behavior
must be rejected. . _

These results clearly contradict those reported by
Maier, and it may be well to conjecture as to the reason for
this. In.Maier's study, open-alley mazes were used. They
were raised, narrow strips of wood connecting various table-
tops, and with this apparatus, extrasmaze cues may have been
important variables. The animals may have been learning a
simple approach response to some extra-maze cue, or to cues

on the open maze itself.

50

In the present study a closed-alley maze was used, with
high walls and a wire mesh tap. Simple approach responses were
minimized by the structure of the maze and by the use of cur-
tains and choice points which were made as similar as possible.
Under these conditions the animals were unable to solve the
problem, and the formation of patterns or gestalten was not
evident.

A further study is suggested by the discrepancy in re-
sults between this study and Meier's study. By repeating this
study, and then running another group of animals in a maze of
identical floor plan, but with raised, open alleys, a compari-
son can.be made between conditions approaching those of minimum
and maximum.condition8 for the learning of Simple approach re-
sponses to gross ones. It may be found that open alleys provide
sufficient cues for the solution of the problem, and Meier's
gestalten.may actually be the learning of simple approach
responses.

Hull predicts problem solution for those groups trained
on the food-water sequence and not for those trained on the
water-food seQuenoe. He discards both a simple association
explanation and Meier's concept of gestalten as emergent new
“wholes.' Hull's analysis is based on the operation of antici-
patory responses as a mediating 84R mechanism which brings about
the solution. According to Hull these mediating S-R mechanisms
are central to the solution of problems and to higher adaptive

behavior in general. He accepts chance as a factor in learning

\ 31
and in problem solution, but contends that chance is tempered
by a behavioral Iloading' of one behavior sequence over another,
which depends upon the animal's store of previous learning.

In the context of the present study, prediction of
problem solution for the Control Group, but not for Group II
follows from Hull's analysis. This prediction is based on
the ”loading“ of the R—U choice over the R-H choice through
the mediation of fractional anticipatory food-taking responses.
These are first elicited in goal box X on the food trials, and
then become associated with the cues of box.U. 0n subsequent
water trials they are elicited by the cues of box U and become
associated with alley I. In the test situation when a hungry
animal is placed at R, with alleys I and III cpen and.alley II
blocked, there will be two excitatory tendencies to alley I.
That is, the cues of the starting box, R, will elicit an approach
response to both alley I and alley III, but the anticipatory re-
sponses elicited by the proprioceptive hunger cues are associated
only with alley I. Thus, the cues of the starting box, plus the
re, constitute two excitory tendencies to allay I. On.the other
hand, there is only one excitatory tendency to alley III which
is produced by the cues of the starting box. Thus the two ex-
citory tendencies to allay I are prepotent over the single
tendency to allay III, and the animal makes the correct choice
and “solves“ the problem. The group trained on water trials

first, and than food trials, will have equal excitatory ten-

32
dencies to both alley I and alloy III on the test trials
because no r3 will have chained back from X to U to R. Thus
with the two responses having equal exictatory tendencies,
the animals in this group should choose one alley equally as
often as the other, and should.do no better than.ahance on
the test trials.

However, the results show that the Control Group did
not solve the problem and did not differ significantly from
Group II on any of the test trials. On the basis of these
results, we must conclude that fractional anticipatory goal
responses did not mediate sufficiently to bring about the
solution. Hull's analysis of the central role of r3 in
problem solving and higher adaptive behavior in general, is
not supported.

It is apparent that Hull intended his analysis to be
applied also to the behavior of organisms higher in the
phylogenetic scale than the white rat. Kendler (3) reports
a study on the Ipferential Behavior lg Preschool Children,
in which he utilized an apparatus analagous to the maze in
this study. From Hull's analysis of the role of r: in
problem solving, Kendler reasoned that inferential behavior
is mediated.by the mechanism of the anticipatory goal re-
sponse. However, his results showed that the sequence of
training most conducive to the successfull mediation of r3

according to Hull's analysis, did not produce a significant

33
amount of inference, as did other training sequences. Thus
Kendler's results also failed to support Hull's hypothesis,
and does cast doubt upon the efficacy of rg as a mediating
representational process.

Of the three groups used in this study, only Group I
made the correct choice with significantly better than chance
success. On each of the three test trials, nine of the ten
animals made the correct choice and ”solved“ the problem.

As eXplained in the section on procedure, this initial food-
water sequence group began to show marked aversive behavior

to goal box H. It was felt that this behavior might con-
stitute an uncontrolled variable Operating in the subsequent
test trials. For this reason, the Control Group--also trained
on the food-water sequence--was added.

Comparison of the results of Group II and the Control
Group shows that the positions of neither Maier nor Hull were
supported, since neither group was able to solve the problem.
However, the success of Group I is consistent with both
theoretical positions, and cannot be used as a test between
the two. The question still remains, then, as to how this
group solved the problem and the other two did not.

Group I and the Control Group were both trained on the
food—water sequence, and ostensibly received the same treat-
ment. However, the aversive behavior to goal box H manifested

by Group I was not evident in the Control Group, and this is

.eaeanu neuade pace 00
use nope-.huon no macaw some new 03%» oumonuon hadec.aega ens .e onwmum o u

 

 

 

 

m r <9
he $.ki ed at he \\ ca 9 .® a. e M r m d . o
f--\ \\,I Imc.¢..\r\aocoha
.v\lttl I.
/\ O.“
W
8. Mn
mu
phi TL
N m
ecu. rD.l.
flunA
can 0 M.
N_nu
GUN
r com S m
./HJ Mn
J... a
8m H.
.2 m
deems 3,.»ch 2:.
HH «~30de I 00%
Se
H Pfiozb

 

35
the primary difference between the two. This aversive be-
havior was first noted on the second day of the water trials
and continued to the sixth day. From the graph of Figure G,
page at, the daily fluctuations in performance can be seen,
and this curve differs considerably from those water acqui-
sition curves of the other two groups. This aversive be-
havior was marked at first by a recoil from the H goal box,
and eventually by a recoil from alley III. It was realized
then that the animals were both hungry and thirsty, being on
twenty-four hour water and food deprivation, and it was sug-
gested that this double deprivation may have been the reason
for the aversive behavior. When the animals were fed to
satiation after the sixth day the aversive behavior ceased.
As seen on the graph, the curve for the remaining days shows
a relatively steady decline and is similar to the curves for
the other groups .

Thus the animals responded negatively to the H goal
box when they were hungry, and positively when they were
thirsty only. On the subsequent test trials, when they were
again put on food deprivation, the negative response to alley
III was again elicited and the animals entered alley I, the
only other possible choice, as a negative response to alley
III. Once in alley I they continued to goal box U, and then
to X and the "solution" of the problem.

36

The Control Group, on the other hand, was well satiated
with food on the water trials, and consequently did not learn
to avoid alley III. In the test trials this group did no
better than chance. It should be pointed out here that the
mean running times differ for the two groups. As can be seen
from Figure 4, page 34, the Control Group performed much
better than Group I, with regard to mean running time and con-
sistency in reduction of running times on subsequent days. It
might be expected, then, that the Control Group learned the maze
much.better than did Group I. Then, consistent with the po-
sitions of both Meier and Hull, this group should have solved
the problem with less difficulty than Group I. The fact that on
the test trials Group I performed.with significantly better than
chance success while the Control Group did not, seems to be fur-
ther evidence that the avoidant responses learned by the first
group were instrumental in the “solution“ of the problem by
that group.

VII. SUMMARY AND CONCLUSIONS

From different theoretical orientations, Meier and
Hull have both analyzed the problem-solving behavior of ‘
white rats. Meier, from a gestalt orientation, explains the
correct choice of an indirect path to a goal when.the direct
path is blocked, as the emergence of new wholes, the insight-
ful reorganization of previous learning into new and successful
wholes. This organization of new patterns is referred to as
gestalten‘by Meier.

Hull, rejecting such ambiguous concepts as insight and
reasoning as not explanatory, analyzed the problem-solving in
terms of fractional anticipatory goal responses. He sees the
"solution” as a behavioral ”loading" of one path over another
through the mediation of re. The ré, according to Hull, is
central for animal problempsolving, and even higher adaptive
behavior. Reasoning or insight is thus reduced to behavioral
chains linked together by the anticipatory responses.

Since neither of the two eXperimental groups set up to
test the two positions were successful in solving the problem,
the predictions stemming from the two analyses were not borne
out. There was neither evidence of insightful solution, nor

of the successful mediation of the r and the two analyses

8’
are rejected.
A third group, for which success was predicted by each

of the two opposing views, did "solve" the problem. But it

38
is felt that the solution came about inadvertently through
the learning of an avoidant response to the incorrect alley,
and not through the formation of gestalten or through the
mediating function of 1-8,

BIBLIOGRAPHY

Hilgard, E. R.: Theories 2;,Learning, Appleton-Century-
Crofts, New York, 1958.

Hull, C. L.: The mechanism of the assembly of behavior
segments in novel combinations suitable for problem
solution. Psychol. Rev. 1935, #2, 219-245.

Kendler, H. H. and Kendler, T. 3.: Inferential behavior
in prezchool children. g, E52, Psychol. 1956, 51,
311- 31 e

Kohler, W.: The Mentality 25 Apes, New York, Harcourt,
Brace , 1925.

Maier, N. R. F.: Reasoning in white rate. 90mg. Psychol.
Monogr. 1929, 6, No. 29.

 

Osgood, C. E.: Method and Theory i3 Experimental Ps -
cholcgy. Oxford Universi y Press, New York, 1935.

Thorndike, E. L.: Agimal Intelligence. New York,
Macmillan, 1911.

I arid;

Demco-293

 

 

N TATE UNIVERSITY LIBRARIES

“III III I I III III I
1 5

M'CIIHIIIII I
3 293 0306] 7 79