AVOMANCE LEAEENENS AND RELEARNING
AS A FUNCTiGN OF SHWTLEEOX WHENSIONS

Thesis {:or fin: Dogma of
EECEEGEXN S'i'éTE Ufii’i’Ef‘
Jay 0. Thomas

1960

M. A.
531“

V
.f
54.

THESIS

 

LIBRARY

Michigan State
University

 

AVOILANCE LEARNIx-SG AM) RtLEARAhG AS A FUNCTION c;

SHUTTLEBOX DIMENSIONS

By
JAY o moms

‘A THESIS

Submitted to the College of Arts and Sciences of
Michigan State University of Agriculture and
Applied Science in partial fulfillment of
the requirements for the degree of

MASTER OF ARTS
Department of Psychology
1960

ABSTRACT

This study investigates the effects of varying
the dimensions; namely, ceiling height and length of
gridway, on the level of original and continued avoidance
learning in a shuttlebox. Seven groups, six eXperimental
and one control, were run under two ceiling heights
(5 anj 14 inches) and three gridway distances (30,26, and 16 inches)
in an original 25 avoidance trials. One-hour later these
groups were given 25 additional retention trials in the
same situation. It was found that tie dimensions of the
shuttlebox, primarily ceiling height, were important in
t‘e learning and relearning of avoidance responses. Animals
in low ceiling boxes initially showed a lower level of learning
but improved when learning was continued following the hour
delay. §s in high ceiling boxes showed higher learning,
but poorer performance when learning was continued following
the hour delay.

A5) Jeff Jr 7;? I ’1’/, / L4 " I
Approved: / w ’"fi<_f¢¢~3l, Date: f Ltk,:t:J/799
Major §$ofessor ‘/ ‘

 

To
Racheal

11

ACKNOflLLDGMmNT

The author wishes to express his gratitude and
sincere thanks for guidance and assistance in tEe planning
and execution of this research, and the develOpement of
this manuscript to Dr. M. Ray Denny, chairman of his committee.
In addition, he wishes to convey thanks to Dr. Charles Hanley
and Dr. Abraham Barch for their excellent criticism and

advice, during the preparation of this thesis.

TABLE CF CCh’ILLnTS

LIST OF TABLES AND FIGURES
L‘ITRODUCTICN
IvItL’I‘lICD
ELSSLTS
DISCUSSION
SUIiE-BRY
APEENDICILS

REFERENCES

Page
vi

\fi

12
15
1b

19

iv

LIST OF lABLLS AND FIGURES
TABLE

Summary of analysis of avoidance responses for
the six experimental groups

Comparison of mean avoidance responses of the
present and the Denny studies

Summarized data

FIGURE

Mean number of avoidance responses in Session I
and Session II

page

11

16
17
18

10

vi

INTRCDCCTION

The shuttlebox has been used in the investigation
of avoidance and escape learning. This apparatus consists
of a rectangular box with a observation window and a grid
floor, either half of which can be charged by the eXperimenter.
When an animal is placed in the situation and given
some consistent signal that temporally precedes the onset
of shock on one-half of the grid, the anihal can learn to
avoid the shock by running to the uncharged grid. The typical
signal is a buzzer or light that comes on five to ten seconds

‘

:‘G

c+

before he shook. Shock avoidance is accomplished by
animal's moving to the opposite end of tTe box when the CS
is sounded. Such behavior is refered to as shuttling. If
the animal does not shuttle the half of the grid upon which
it is situated is electrically charted and continues to be
charged until the animal moves to the uncharged region of
the alley.

dhen an animal shuttles following the onset of the
CS but belore shock, it is said to "avoid". If it shuttles

"escape". The occurrence

after shock begins, it is said to
of an avoidance reSponse prevents both escape behavior and
the eXperience of being shocked. In an escape study, the
animal is shocked on each trial; the present investigation
deals with avoidance behavior.

Few studies have been conducted on t‘e retention

of an avoidance resgonse; probably because from the available

literature, it a;[eared long lastin; and hi hlv resistent

to extinction. Recently, however, Kamin (1957) has investirat=d
the retention of an inconpletely learned avoidhnce resgonse.

If original learning were interrupted scsn enouyh, he reasoned,
only partial retention would occur, and a retention curve

could be plotted.

Kanin ran hooded rats for 25 trials in a typical
shuttlebox and measured retention in 25 additionsl trials
following celars of O, 1/2, 6, and 24 hours and 19 days.
The results were un xpected; instead of a monotonic decreasing
relation of retention to time, he found a V-shaped curve
which declined from C to one hour and then rose from one-
hour to 24 hours. Differences in retention at C, 24 hours,
and 19 days were not statistically significant.

Kamin interpretated his data in terms of two
independent processes; one for each sebnent of the curve.

The first segment of the curve, 0 to one hour agrees with
tie vernacular concept of forgetting. The rising segmentv
of the curve represents an incubation effect, a jelling of
the avoidance habit following the initial decrement in retention.

Denny (1958) reinterpreted Kamin's V-shaped curve
(the "Kanin Effect") in terms of the incubation of anxiety
rather than the incubation of an avoidance habit. according
to this interpretation, anxiety initially builds up in the
interval immediately following the original learninL trials
to a point where it interferes with the act of shuttling.

As observed by Denny, animals when tested one hour later,

typically freeze in a second session, and this behavior is

P.)

incompatable with shuttling. Following a delay of approximately
one hour the anxiety begins to dissipate and retention of

the avoidance reSpcnse is clearly apparent after 2 hours.

From this point of view, it was predicted that if the anxiety
could be kept from building up, the § would no longer show

a decrement in performance following an hour delay.

USinL delays of O, 1, and 24 Fours Denny employed
the following methods to inhibit the prowth of anxiety during
t“e one-hour delay; 1) gs were on a reduced feeding regimen
for two weeks and under 24—hours food deprivation when trained.
They were allowed to eat alone in the home cage during the
hour interval (Counter conditioning), 2) §s were left in the
shuttlebox during the hour interval, without shock or buzzer
present (Desensitization). It was also predicted that if the
same amount of shock as the § received in original learning,
were readministered in a different situation on the follow-
ing day, the incubation of anxiety would be reinstated. Thus,
if a second set of 2S learning trials followed one-hour later,
the “Kamin Effect" would appear, even though it was now 24
hours since original learning.

All these manipulations were carried out, and all
yielded results which supported the incubation-of-anxiety
hypothesis.

However, there were certain differences in the
results between the two studies; namely, considerably better
original learning and slightly better 'Kamin Effect” after a
one-hour delay in Denny's study as compared with Kanin's
(a mean of 10.7 vs. 5.7 avoidances in Session I and 10.1 vs.

6.6 avoidances in Session II, respectively). Concomitantly

there were certain apparatus difyerences: Denny's shuttlebox

had a higher ceiling (14 vs. 4 3/4 inches) and a s orter

Lridway (26 vs. 36 inches). Although other possible method-
olOgical differences may have obtained between the two studies--
the level of shock, for instance, or the location of the
observation window—- the apparatus dimensions appeared most
relevant. Thus the present study is an investigation of the
effect of length and 1eight of shuttlebox, both on oriLinal

learning and relearning one-hour later.

s
I

The §s were seventy experimentally-naive nocded,

black, and albino rats from the colony of the Department of
Psychology at Michigan State University. The 35 m“le and 35

female rats ranged in age from 90 to 150 days. During the

study they were maintained on an ad lib feeding schedule and
housed five gs of the sane sex per cage. Animals were randomly
assigned to seven groups, except for balancing the nunber

:h
l

of males and enales in each group. Running order of experimental

and control 'rcups was also randomized.

A C.J. Applegate stinulator was used to provide a
continuous shock of 1.7 millianps directly to the Grid. The
full distance of the gridway or either half could be charged,
but only one half was charged at any one time. The §s completed
the circuk by making contact wit“ any two of the adjacent COpper
rods which were set approximately 1/4 inch apart in the grid

floor. The shuttlebox, painted flat blacx with a glass front,
was desiLned with removable partitions so that it could be

any of three lengths: 36, 26, or 16 inches. The fixed wood
ceiling was 14 inches high. To ef ectively lower ceiling
height a sheet of clear glass was inserted five inches from

tEe floor, thus maintaining visual similarity for the two

hei¢hts. These modifications of the shuttlebox provided six
combinations of apparatus conditions:

1) 36 inch long gridway with a 14 inch ceilin;

2) 2c inch long gridway with a 14 inch ceiling

3) 16 inch long gridway with a 14 inch ceiling

4) 36 inch long gridway with a 5 inch ceiling

5 26 inch long gridway with a 5 inch ceiling

6 16 inch long gridway with a 5 inch ceiling

cl! «I111 urn]

One experimental Croup of S5 was run under each set of conditions.

Thus the present study includes the Kanin (36 inch
long bridway with a 14 inch ceiling) and the Denny (26 inch
long Lridway with a 5 inch ceiling) conditions. The width of
all three shuttleboxes was four inches. The only difference
between Kamin's shuttlebox and the conprable conditions of the
present study is that the former had a glass tOp wfiile the
present apparatus had a glass front. The conditioned stimulus
was a buzzer of seventy decibels activated by six dry cells.

The study was divided into two sessions: an inconplete
learning phase of 25 avoidance learning trials (Session I)
and a block of 25 additional trials (Session II), which came
one-hour after the conpletion of Session I.

In Session I S was placed in the shuttlebox for
one minute before the trials began. The buzzer (CS) was
sounded for five seconds prior to shock (US) and terminated
when S crossed the midline of the box, either by avoiding or
escaping. When § failed to avoid, the CS and US overlapped
and both were response terminated. The inter—trial interval
was one minute. Occasionally on the first trial, an S would
shuttle to the CS alone; in such cases this trial was not
counted as one of the 25, and the buzzer was again sounded
following the inter—trial interval. In ot er words, t‘e
initial trial was the first trial on which tbe animal did not
respond to the CS alone and took shock.

At the termination of Session I, S was returned

to an outer room to be placed in its livinc cage with its

([1

cage mates present. One hour later S was returned to th

apparatus for 25 more trials (Session II). For any one

1m

5

the conditions 01 See ion II we e identical wi h t‘e conditions

(‘1

of Session I.

A control proud was run in a shuttlebox 26 inches

with a 14 incL ceiling with Session II innediately followinb
Session I. In other words, this group had 50 continuous
avoidance trials. The control was included primarily to
deternine whether t”e present sanple of rats would perform in

the same manner as in tTe earlier Denny study. For the main

purpose of the present study the control group was unessential.

RESULTS

The eXperimental desipn, excluding the control
group, lent itself to a 2x2x3x2 analysis of variance with
repeated measures on §S. This design was taken from Lindquist
(1956) and is a type III design with one added factor. The
analysis is summarized in Table I. Two main effects-~ceiling
height and sessions-~and three two-way interactions were

significant. Ceiling height was significant at the .05 level

J)

an sessions at the .001 level. None of the third order or
the fourth order interactions were statistiCally significant.
A plot of the mean number of avoidances for Session I and
Session II of the six eXperimental and the control Lroup is
presented in Figure I. The neans of the echrinental proups
were compared by a Studentized Distribution and yielded the
following significant differences: 1) SS in high ceilinL boxes
made more avoidances in original learning (Session I) than
did SS in low ceiling boxes, 2) SS in low ceiling boxes sVowed
improvement (i.e. the nean number of avoidances in Session II
was greater than in Session I), w*ereas §s in high ceiling
boxes did not s?ow inprovenent, 3) performance in Session II
in a low ceiling short (16 inch) box was better t‘an performance
in Session II in a high ceiling short (16 inch) box.

The significant interaction between height and
length (distance) tells us that with a high ceiling learning
tends to be better the longer the alley and t”at with a low

4‘5.

ceiling just the reverse is true The sipnilicant interaction

T

ABLE

1

Analysis of Avoidance ReSQOnses

Source of Variation
Between

Sex

Height

Distance

Sex x Height

Height x Distance
Sex x Distance

Sex x Dist x Height
Error

Within

Sessions

Sessions x Sex
Sessions x Height
Sessions x Distance
Sess x Sex x Height
Sess x Sex x Distance
Sess x Height x Dist
Sess x Sex x Dist x H
Error

Total

* 4 .05
M 4 .01

DF

48
60

Mean Square

\0

4.66*

7.52**

c.71*

187.5**

450.0*«

IL/

MEAN NUMBER OF AVOIDANCES

N04.

 

 

 

 

 

 

 

 

D
XIIIIX mem_OZ _
OIIIIO Mm mw_OZ N
a D nozamorcaxmmv
_ T
O X
/ O
.F / o
.o 0 X1 )X\ D
O O
Ui- X/ XIX
o _ + a a L. w .||_
mx_m mxmm mxwm 3X5 .Axmm £wa Exmm

CORT

ERR

between sessions and hei;ht indicates, at a minimum, that
the "Kamin Effect" occurs to a greater extent with a hi¢h
ceiling than with a low ceiling. Finally the significant

interaction between sex and ceiling height occurs because

females made considerably more avoidances in the hith than in

the low ceiling box and males made a few more avoidances in

the low Ceiling box than in the high. The control group

performed in both sessions at levels comparable to Denny's

controls, and perfortance during relearning was excellent.

11

From Session I to Session II, the control group showed significant

improvement, both when conpared with itself and with the
26 inch x 14 inch eXperimental group (see Table 2).
TABLE 2

Comparison of Mean Avoidances of
Present and Denny Studies

EXperimental conditions Session I Session II Improvement

 

 

Present 14 x 26 inches 9.6 18.8 +9.2
Denny 14 x 26 inches 10.2 17.8 +7.6
Present 14 x 26 inches 9.5 8.2 -1.3

 

Denny 14 x 26 inches 10.7 10.1 —0.6

 

DISCUSSION

The present investigation appears to resolve the
major discrepancies between the Kamin and Denny studies.
Under comparable conditions conparable results were obtained.
The only possible exception was the finding or less "Kamin Effect“
under conditions similar to Kamin's. Almost dramatically,
avoidance learning and relearning was affected by the dimensions
of the shuttlebox used.

The superiority in original learning of high ceiling
vs. low ceiling sugwests that the low ceiling limits the
number of avoidance-type responses in the animals repertoire,
and this in turn promotes the non-instrumental act of freezing.
The effect is to decrease the probability of making the
final shuttling reSponse. The experimenters observation of
animals in low ceiling groups frequently freezing, with the
onset of buzzer, would tend to support such an interpretation.

If the § in a low ceiling box is to learn the
avoidance reSponse, freezing behavior must extinguish. This
may account for the relatively better performance after one-
hour delay for the low ceiling §S. The one-hour delay, according
to the incubation-of-anxiety hypothesis would, in general,
increase the tendency to freeze. SS in high ceiling boxes,
who have not been freezing, have not had an Opportunity to
extingu sh this type of reSponse and consequently are at a
disadvantage, i.e. show more "Kamin Effect".

Conversely, §s under high ceiling conditions can

make more reSponses to the stimulus situation: jumping,
standing, and leaping. These responses can readily chain in
with the correct response of shuttling. At first ;lanee,
these results pose a paradox, for by lowerin; the ceiling

one might expect an increase in the Specification of the
correct response of shuttling, instead, it appears that we
are promoting a reSponse which is inconpatible with shuttling
and inconpatible with responses allied with shuttling. Such
an interpretation.is consistent with new theoretical Behavioristic
approaches that emphasize the importance of each reSponse
elicited at every step of the analysis (Denny and Adelman,
and Logan).

The less important variable of length may Operate
as follows. Under low ceiling conditions, the escaping rat
can only run, which means it runs through shock to escape
shock, being consistently punished during the early stages
of making the correct response. Thus learning tends to be
poorer the longer the box. In high ceiling boxes, where
escape is possible in a number of ways, this length factor
is not as critical.

The interaction between sex and ceiling height
indicates that females made more avoidance responses under
high ceiling conditions than under low ceiling conditions.
Males, it was found, avoided slightly less under low ceiling
_conditiens. In order to shed some lith on this interaction,
it should be mentioned that Kamin, with a low ceiling box
and 1.1 millianp shook level, found no differences between
males and females (personel comnunication to M. Ray Denny),

whereas Denny found it necessary to use a higher shock

14

level with females in order to Let the "Kamin Effect? (1.7
milliamps for females vs. 1.1 milliamps for males) with a

high ceiling. From this fact, it can be inferred that females
are particularly prone to freeze in a low ceiling box but

will freeze in a high ceiling box only if the shock level is
sufficiently high. Therefore, it is not sumrising with the
shock level of the present study that females froze more and
thus avoided less under low ceiling conditions than was the

case under high ceiling conditions.

15

SUMMARY

This eXperiment studied the effects of shuttlebox
dimensions upon the amount of avoidance learning and relearning.
Seventy hooded, black, and albino rats--35 male and 35 females--
were given 25 shuttlebox trials and then rerun one-hour later
for an additional 25 trials. It was found that tie dimensions
of the shuttlebox, primarily ceiling height, were important
in the learning and relearning of avoidance responses. Animals
in low ceiling boxes initially showed a lower level of learning
but improved when learning was continued following the hour
delay. gs in high ceiling boxes showed higher learning, but
poorer performance when learning was continued following the
hour delay. A tenative interpretation of the results was
included, which seemed consistent with new behavioristic

positions.

APPENDICIES

SUMMARIZED gain

 

 

 

 

16

fiiofg F Session I. Session II

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

Blocks of

5 trials
1 C 2 C 1 1 O O 1 O 1 6 O 1 O 3 1 2 1 1 2 O 11
2 1 1 1 1 O 1 O 1 1 O 7 3 1 1 O 1 O 2 O 1 3 9
3 2 3 3 C 2 4 2 2 3 2 23 C 3 3 1 3 2 1 2 3 1 19
4 3311222244 24 4111142221 19
5 3 3 5 5 3 2 4 3 2 1 31 3 5 3 C 1 O 2 4 2 4 24

Group

14 x 26

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

Blocks of

5 trials
1 C O O 1 O O O O O O 1 O 1 1 1 O 1 5 O 4 O 13
2 1 1 3 O 2 O 3 O 2 1 13 C 1 2 C 1 2 4 1 4 3 18
3 O 5 3 1 1 4 5 1 5 2 27 O 4 3 1 3 3 2 O 3 O 19
4 1 3 5 1 1 4 5 O 3 1 24 4 O 5 O O 2 2 1 O 1 15
5 1 4 4 O 2 4 5 5 4 1 26 1 2 4 1 1 3 3 O 1 2 18

Group

14 x 36

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

Blocks of

5 trials
1 1 O 1 O 1 1 1 O 1 1 7 2 O O 3 4 2 1 1 3 O 16
2 1 3 2 C 3 4 2 4 1 3 23 1 2 1 3 1 2 2 2 2 4 2C
3 2 2 1 1 2 2 3 4 3 A 24 2 2 3 1 O 1 2 2 3 3 19
4 3 2 3 5 4 3 5 4 5 1 35 3 2 1 2 3 2 4 5 4 1 27

17

 

 

 

 

groapé Session_£ Session II

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

Blocks of

5 trials
1 O 1 O 1 C O O 1 O O 3 1 1 3 1 2 O O 2 2 2 14
2 1 O 1 1 1 1 O O O 1 b 2 1 1 3 1 2 O 2 1 4 17
3 3 2 1 1 3 1 O O 1 1 13 4 4 2 2 4 4 2 3 2 3 3C
4 2 O 2 2 3 3 O 1 O 1 14 4 4 4 5 5 3 1 4 3 3 36
5 3 1 3 2 4 2 1 3 O 1 40 5 2 3 4 5 3 3 3 3 3 34

Group

5 X 26

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

Blocks of

5 trials
1 1 C O C O 1 O O O O 2 1 O 1 O 2 O O 4 3 O 11
2 C C O 1 O O 1 1 1 1 5 3 O 2 3 1 1 2 O 1 1 14
3 2 1 1 3 2 O 1.3 2 1 16 O 4 2 1 3 2 1 3 5 4 25
4 1 2 1 1 O O 2 O 2 O 9 4 5 4 3 4 2 4 3 3 4 36
5 2 4 3 1 3 1 3 1 1 1 2O 5 4 5 3 2 1 5 4 4 2 35

Group

5 x 36

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

BlocKs of

5 trials
1 C O O O O 1 O 1 O O 2 O O 1 2 1 2 O 1 O O 7
2 O 1 1 1 O O 1 1 1 O 6 O 3 2 3 3 4 2 1 2 0 2O
3 1 1 O 2 1 2 O 1 2 O 10 3 O 1 2 2 1 2 3 1 2 17
4 2 1 2 1 1 3 1 1 O O '12 4 2 2 1 1 5 3 3 O 1 22
5 O 1 2 2 2 3 1 3 1 2 17 2 2 4 5 5 3 2 2 3 3 31

 

 

17

 

 

Group Session I Session II

5 x 16

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

Blocks of

5 trials
1 O 1 O 1 O O O 1 O O 3 1 1 3 1 2 O O 2 2 2 14
2 1 O 1 1 1 1 O O O 1 6 2 1 1 3 1 2 O 2 1 4 17
3 3 2 1 1 3 1 O O 1 1 13 4 4 2 2 4 4 2 3 2 3 3C
4 2 O 2 2 3 3 O 1 O 1 14 4 4 4 5 5 3 1 4 3 3 36
5 3 1 3 2 4 2 1 3 0 1 20 5 2 3 4 5 3 3 3 3 3 34

Group

5 x 26

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

Blocks of

5 trials
1 1 C 0 0'0 1 O O O O 2 1 O 1 O 2 O O 4 3 O 11
2 C C O 1 O O 1 1 1 1 5 3 O 2 3 1 1 2 O 1 1 14
3 2 1 1 3 2 O 1 3 2 1 16 O 4 2 1 3 2 1 3 5 4 25
4 1 2 1 1 O O 2 O 2 O 9 4 5 4 3 4 2 4 3 3 4 36
5 2 4 3 1 3 1 3 1 1 1 2O 5 4 5 3 2 1 5 4 4 2 35

Group

5 x 36

Subjects 1 2 3 4 5 6 7.8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

BlocKs of

5 trials
1 C O O C O 1 O 1 O O 2 O O 1 2 1 2 O 1 O O 7
2 O 1 1 1 O O 1 1 1 O 6 O 3 2 3 3 4 2 1 2 0 2O
3 1 1 O 2 1 2 O 1 2 O 10 3 O 1 2 2 1 2 3 1 2 17
4 2 1 2 1 1 3 1 1 O O 12 4 2 2 1 1 5 3 3 O 1 22
5 O 1 2 2 2 3 1 3 1 2 17 2 2 4 5 5 3 2 2 3 3 31

Group Session I Session II
Control

 

 

Subjects 1 2 3 4 5 6 7 8 9 10 Total 1 2 3 4 5 6 7 8 9 10 Total

Blocxs of

5 trials
1 1 O O O 1 C C 1 O 0 3 4 5 4 5 5 3 2 4 4 O 36
2 1 3 2 2 1 O O O 1 2 12 2 2 3 4 5 4 5 5 2 1 33
3 5 2 5 3 4 1 O 1 2 O 23 5 4 5 2 5 5 4 4 5.2 41
4 3235341240 27 5444345231 35
5 4 3 3 4 4 5 1 2 4 1 31 4 5 4 5 4 5 5 4 5 2 43

19

REEERENCpS

Denny, M.R., and Adelman, H.M. Elicitation theory: An
analysis of two typical learning situation. Psychol. Rev.,
1955, 02, 290-296.

 

Denny, M.R. The "Kamin effect" in avoidance conditioning.
Amer. Psychol., 1958, 13, 419.

Kamin, L.J. The retention of an Incompletely learned
Avoidance Response. J. comp. physiol. Psychol., 1957, 50,
457-460.

Logan, F.A. .A Micromolar Approach to Behavior Theory.
Psychol. Rev., 1956. 63. 03-73.

Lindquist, E.F. Design and analysis of EXperiments in
Psychology and Education. Boston Houghton Mifflin, 1956.