OVEELEAB.’EE‘% REVERSAL EN 343
EN SFAHAL ESSCEEMEM'E’ESNS

T210222 *369 S‘éw $013990 of M. A.
MICHEGAN STATE WHERE???
Daniel Francis Tortora

12970

THESIS

f L [3 it A If.
hiic}dgaxl State

University

   
   

 

 

a;
t;

';4-o-w.—-_~ vu— _ -
~0~ ”w“. .

 

 

Lm ' "105

ABSTRACT

OVERLEARNING BEVEBSAL IN RATS
IN SPATIAL DISCRIMINATIONS

By Daniel Francis Tortora

The overlearning reversal effect is defined as the ‘
difference between two groups On the acquisition of a
reversal task. One group is given overtraining after
original acquisition and then is reversed, the other group
is reversed immediately after original acquisition. When
overtraining facilitates reversal, ORE is positive, when
overtraining retards reversal, ORE is negative. Two
opposing views on the nature of discrimination learning have
been examined. One model, that of Mackintosh, hypothesizes
that discrimination learning proceeds in two stages. First,
the animal learns to pay attention to the relevant dimension(s)
and ignore irrelevant dimensions. Second, the animal learns
to respond apprOpriately to the correct cue. It is
postulated that switching to relevant dimensions and
switching away from irrelevant dimensions takes a consideribly
long period of time. Thus, the reversal of any task
(using large reward) that includes many irrelevant stimuli
will be facilitated by the extra practice afforded by over-
training. The other model, the elicitation position of
Denny, and the one guiding this thesis postulates a single-

stage analysis of discrimination learning, in which the laws

Of classical conditioning are exploited. CS-US contiguity
is the necessary condition for learning. *The elicitation
position states that for a complete understanding of ORE
one must perform a detailed analysis of the CS-US relations
arranged by the experimenter. Such an analysis is pre-
sented in the body of the thesis. This study explored ORE
in spatial discriminations in which the CS-US relations
were controlled; five experiments were performed.

The first three experiments explored the effect of
overtraining upon the subsequent reversal of a kinesthet-
ically controlled turning response where size of incentive
and length of the stem of a T-maze were manipulated. It
was found that where the stem of the maze was long, as in
experiments I and III, a significant negative or reverse
ORE was obtained. However, when the stem was short, as in
eXperiment II, no ORE was obtained.. It was suggested that
size reward was inversly related to the magnitude of the
negative ORE. Experiment I, using a small reward, yielded
a greater negative ORE then experiment III using a large
reward.

Experiments IV and V investigated the effect of over-
training on the reversal of a place response in which all
controlling stimuli were relevant and redundant; two different.
reward sizes and a low acquisition criterion were used.

A significant negativé ORE was found for the small reward
groups in experiments IV and V. A significant positive ORE

was found for the large reward group of experiment V.

OVERLEARUING REVERSAL IN RATS

IN SPATIAL DISCRIMIRATIOUS
By

DANIEL FRAKCIS TORTOHA

A THESIS

Submitted to
Hichigan State University
in partial fulfilment of the requirements
for the degree of

". NOW”? 1'1 am
finals.“ l‘ An. ls

Department of Psychology

1970

Approved: kflf/ fiwf‘

 

.....

ACKNOWLEDGEMENTS

I would like to acknowledge the help given to me dur-
ing the course of this research. I would particularly
like to thank Dr. M. Ray Denny for the guidance and support
he has given throughout my endeavors.

I would like to thank Dr. S. Ratner for his inspira-
tional and thought provoking comments concerning behavior
modification and experimental validity and Dr. L. Hyman
for his incisive critique of the thesis.

I would also like to thank my wife, Angela, who Spent
a good many hours proofreading and typing this thesis, and

for her unfailing moral support.

ii

TABLE OF COHTEN S

Page
ACKNOWLEDGEMENTS..................................... ii
LIST OF TABLES....................................... V
LIST OF FIGURES...................................... vii
INTRODUCTION......................................... 1

PBOBLEFJIO0.00.00...OOCOOOOOOOOOOOOOOOOOOOOOOOOOCOOOOCO 15

EXPERIMENT I......................................... 16
Introduction.................................... 16
Method.......................................... 16

Subjects................................... 16

Apparatus.OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO. 16

Procedure. 0 O O O O O O O O O O O O O O O O O O O O O O O O O 0 I O O O O 0 1?
BesultSOOOOOOOOOOCOOOOOOOIOOOOOOOOOOOOOOOO0.0... 20
DiSCuSSiOlfl-I O O O I O O O O O O O O O O I O O O O O O O O O O O O O O O O O O O O O O 21+

EXPEBII'IEIIT IIOOOOOO000.00.00.00...OOOOOOOOOOOOOOOOOOO 26

IntrOdUCtion. O O O O O O O O O O C O O O O O O O O O O O O O O O O O O O O O O O O 26
DilethOd-OOOOOOOOOOOOOO0..O...OOOOOOOOOOOOOOOOOOOOO 27
subjeCtSO O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O 27

Apparatus.................................. 27
Procedure.................................. 27
Results......................................... 28
Discussion...................................... 30
EXPERIMENT III....................................... 31

IntrOduCtj-OHOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO 31

iii

Page

I'IethOd-OOOOOO0......COOOOOOOOOOOOOOOOOOO0...... 32

\

Subjects................................. 32
Apparatus................................ 32
Procedure................................ 32
Results....................................... 34
Discussion.................................... 38
CONCLUSION FOR EXPERIMENTS I, II, III.............. 39
IN RODUCTION FOR EXPERIMENTS IV, V................. 41
EXPERIEELT IV...................................... 42
Method........................................ 42
Subjects................................. 42
Apparatus................................ 42
Procedure................................ 42
Results....................................... 44

HIJERIIIEIIT V.o.o.000000000.oooooooooooooooooooooooo [+7

IlltrOduc-tion. O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O 0 L47
IIethOd-O OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO 47
subjeCtS O 0 I O O O O O O O O O O O O O O O O O 0 O O O O O O O O O O O 0 LL?

Apparatus................................ 47
Procedure................................ 47
Results....................................... 49
DISCUSSION FOR EXPERIMENTS IV, V................... 53
GENERAL SUMMARY OF RESULTS & COECLUSIOES........... 57
REFERENCES......................................... 63

APPEI‘IDICESOOOOOOOO...CcOOOOOOOOOOOOOOOOIOOOO0...... 68

iv

10.

ll.

12.

13.

LIST OF TABLES

Results of Reversal Experiments with rats.....

An analysis of Overlearning Reversal Effect
(01-1E)inratSOOOOOOOO0.0.0.0000...0.0.0....00

Mean trials to criterion for original
acquisition and reversal for the experimental
and control groups in Experiment I...........

Number of subjects in each group exhibiting
a specific position preference in Experiment I

Summary of the analysis of Covariance with
trials to criterion as the measure in
EXpediment IOOOOOOOOOO...OOOOOOOOOOOOOOO0.0..

Mean trials to criterion for acquisition and
reversal of experimental and control groups
in EXpeI’iment 1100.00.00.00.0.000000000000000

Number of S's in each group exhibiting a
specific position preference in Experiment II

Summary of the Analysis of Covariance in
Experiment IIOOOOOOOOOOOOOOOOOOOOOOOOO0.0.0..

The mean trials to a criterion for experimental
and control groups in Experiment III.........

Number of subjects demonstrating position
preference in Experiment III.................

Summary of the Analysis of Variance using
trials to a reversal criterion in
EXperimel’lt IIIOOOOOOOOOOOO00.000000000000000.

Mean and variance for the four groups used
in the anovar summarized in Table 11 in
Experiment IIIOOOOOOOOOO0.0000000000000000000

Summary of the acquisition data for two
measures and both groups in Experiment IV....

Page

20

21

28

29

29

36

37

44

Table
14.

15.

16.

17.

Summary of the reversal data for three
measures and for both groups in
EXpePiment IVOOOOOOOOOOOOOOOO...0000.00.00.00

a) Trials to criterion during reversal for
all four groups in Experiment V..............

b) Trials to Last Error during reversal for
four groups in Experiment V..................

Summary table for the Analysis of Variance
with trials to the last error as a measure
in EXperimen-t VOOOOOOOOOOOOOOOOOOIOO000......

Summary of all results of reversal learning
from all experimentsOOOOOOOOOOOOOOOOOOOOOOOOO

vi

Page

45

49

50

51

55

LIST OF FIGURES

Figures Page

1. A schematic representation of the
arrangement of the running room and the
T-maze used in all experiments...........16—17

2. Graphic presentation of the interaction

of incentive size and amount of training
from EXperj-rfiel’lt .VOOOOOOOOOOOOOOOOOOOOO... 52

vii

INTRODUCTION

The Overlearning Reversal Effect (ORE) can be defined
as the facilitation of reversal learning as a result of
overtraining. This phenomenon is usually investigated
using a two group eXperiment. Both groups are trained to
a criterion on a discrimination task, eg. 18 correct out of
20 consecutive responses. The eXperimental group is then
given from 100-300 percent overtraining on the original
task and then reversed. The control group is reversed
immediately after it has reached Criterion on the original
acquisition. During reversal, the cue values are reversed,
i.e9 the stimulus that previously led to reinforcement now
leads to non-reinforcement, and the former negative stimulus
is now positive.

In 1953, Reid performed the first overlearning reversal
study. He found, contrary to theoretical preconceptions,
that overlearning facilitated reversal. In the next 16 years
over 50 studies had been performed and several theoretical
analyses (Lovejoy, 1966, 1968; Mackintosh, 1965a, 1969; Paul
1965; and Sperling, 1965a & b) have been postulated to
explain this phenomenon.

Table 1 presents the results of such experiments. An
undeniable conclusion can be gleaned from a perusal of these
studies. Up to the present there is not one hypothesis or

theory which can explain the divergent results.

 

2 Acmeafiv amoegaxeaa
Ammaev aawaom s ammeeeseeaa
Ameaav gopamao

Ameaav consume a .AEeam .Haam
Anneafiv emeam e Haas

Ammmfiv pmmmoam @ moamse
Ameafiv panama s exam
Aeneaav amopeaxemz
Amemfiv “negox

Amwaav as am Hana
Aamaav aasm e Lepesaeo
Ameaev aeaomneaam
Ameafiv neasm a oamea_m
Ameaav meowme a oaee<.a
Ap.emeaav gossaao

Ammmav maaaaeam

Amwaav nemmoam s moaese

Ammaav seem

Ammafiv maonza

Ammafiv soaaaflo s apnea

gamma .pmeaa .Qmeaa .Neaav emoeaaaoma
Aamaav flanges

Aaeaav soaam a nomH

‘ Ammmfiv pogoom

Ammaav somaeeeem s definaao

Ameaev aeaeamo

Awmmfiv somfiamz s 6309.0 .pmadsmz .sosdsm

Aaomfiv Hamm a .Qmmpmz .OLHwaoopm

 

Hamnm>mm
mdepom wsfiqfimppno>o

Hmmpm>om so pommmm
02 ms: mnflsamnpao>o

Hmmao>om
mmpmpHHHOMQ wsfisflmppno>o

 

 

.mpmm sum; mucosapogxm Hmmao>mm mo madamomuu.fi mqm<e

From Table 1, it can be seen that this proceedure has
yielded three kinds of experimental results. They are as
follows: (a) the experimental group learns to reverse faster
than the control group-ORE or positive ORE, (b) there is no
significant difference between the two groups-no ORE, and
(c) the control group reverses faster than the experimental
group-reverse or negative ORE. Thus any satisfactory
explanation of ORE must also explain its variants.

ORE is studied either in a visual discrimination task
(brightness or pattern discrimination) in which values of the
positive and negative cues are reversed or in a spatial dis-
crimination task (T-maze or y-maze) in which the direction
of turn is reversed. When a large incentive has been used
in the experiment, ORE occurs equally often in visual and
spatial discriminations.

Another important aspect of ORE is that it is a robust'
phenomenon. When differences do occur they are usually
very substantial, necessitating only a few animals to obtain
significant results. What is in question here is not in-
ternal but external validity. When, where and how it occurs
are the pressing issues. This can only be accomplished by a
thorough analysis of the procedural differences among
diverse experiments in order to identify the functional

variables.

THEORETICAL ANALYSIS OF ORE

Of the theories that have been proposed to answer the
above questions, two theories of learning, the two stage
discrimination model of Mackintosh (1969) and the elici-

tation position of Denny (1967, 1970) will be reviewed here.

Mackintosh's Theory of ORE

Mackintosh postulates a two-stage attentional model to
explain ORE (Lovejoy, 1966). The first stage is the learn-
ing of the apprOpriate stimulus analyzer. Here § learns
which stimulus dimension to attend to and which dimensions
are irrelevant. It is postulated that learning not to
attend to irrelevant dimensions is a relatively slow pro-
cess taking much longer than learning to attend to the
correct dimension and make the correct response. The
second stage is the learning of the cue value of the
relevant stimulus dimension. Once this is accomplished
discrimination is said to be complete.

For Mackintosh, the occurrence of ORE depends on the
joint satisfaction of two conditions-a difficult discrim-
ination and a large incentive. He assumes that ORE occurs
only when the probability of attending to the relevant
dimension is not very high at the outset of training.
Another way of saying this is that the ORE will only occur

when rats are trained in a relatively difficult discrimination.

5

In such a situation, overtraining increases the probability
of attending to the relevant dimension and decreases the
probability of attending to the irrelevant dimension(s).
Thus overlearning facilitates reversal learning. The
animal has only to learn the new cue values since he is
already attending to the correct dimension and ignoring the
irrelevant ones. The immediate reversal group has not
learned to ignore irrelevant dimensions. Thus, during
reversal, it shifts from dimension to dimension until the
correct one is rediscovered. Only then can it proceed to
learn the reversal discrimination. When the discrimination
is easy the relevant dimension is strong from the outset
(i.e. strong in relation to the irrelevant dimensions),
and no benefit is supposed to accrue from overtraining.
Mackintosh introduces size of incentive into his theory
by assigning parameter values to 9reward" operators in the
Bush and Mosteller stochastic model of learning. The actual
psychological significance of magnitude of reward is not
postulated.

Elicitation Theory

 

This theory is monistic in its approach to learning.
All learning is viewed in a contiguity framework in which
classical conditioning principles are exploited. The theory
states that learning depends upon consistently eliciting
the to-be-learned response in close temporal contiguity with
a particular stimulus situation. This in turn means
minimizing the elicitation of alternative responses to the

same or similar stimulus situations. In this way, the

response that is being learned wins out over other possible
responses. The function of incentives or reinforcers is
that they are important elicitors in the learning situation.
These stimuli elicit a characteristic response over and
over again and thereby minimize the occurrence of competing
responses.

Elicitation theory states that the UR of approaching
the goal object (foOd, water or whatever) is always present
in an instrumental learning situation and that this UR
mediates the learning of the entire instrumental chain.
Eventually the animal learns to approach the goal object via
this chain.

According to the theory the absence of the incentive
in the previously reinforced goal area is itself a US which
elicits antagonistic responses that take the animal directly
away from the goal area. The frustration-instigated
antagonistic responses consist of aggressive or tangential
responses,i.e.,very possibly displacement activity when S
cannot actually escape from the frustrating situation.

Denny postulates that during discrimination learning both
the response class of approaching the goal object in the
context of the positive cue and withdrawal from non-
reinforcement in the context of the negative cue are learned.
Elicitation Theory Applied to ORE

According to elicitation theory, (Denny, 1970) reversal
learning is accomplished by the extinction of the original
response. Extinction is considered to be the result of
counterconditioning. It is postulated that without the

competition from an alternative response the original response

7

will never be replaced. Secondary elicitation is the mech—
anism used to handle this phenomenon. By this is meant that
extinction is due to the elicitation of withdrawal by the
ommission of the food from the food tray. The elicitation

here is called secondary elicitation since the animal must

have experienced food at the food cup, in the presence of the
prevailing stimuli, in order for the absence of food to be

an elicitor. Thus the removal of food serves as a unconditioned
stimulus, consistently eliciting a characteristic class of
antagonistic reSponses (RC). Rc's are conditioned to con-
tiguous stimuli and through backchaining becomes conditioned

to earlier stimuli in the chain. Eventually RC competes effec-
tively with all stages of the original instrumental response
(Ro). In reversal learning, this means that the Rc must occur
at the choice point area before it affects the choice reSponse.
This occurs via backchaining which is initiated at the empty
food cup in the context of whatever stimuli are presented at
the time of non-reinforcement.

In order to explicate the elicitation position of ORE
further, an analysis of current research will be undertaken.
Table 2 presents a summary of research methods used and
results obtained in this area. It must be pointed out here
that the organization of the table is based upon the con-
viction that a complete understanding of ORE can only be
accomplished by a detailed analysis of the procedures used
by the various experimenters. Elicitation theory helped

in abstracting the critical stimulus-response relationships.

 

meme .meHHeoam
meme .am0pnaxeez

meme .eaepeeam

Home .Hamm
samgpmz .mefinmxoopm
meme .emoaseaomz

meme .eeaem s o»me<.a
meme .emeompeaam

wwmfi .mxmzmmmmxsq
mama .em0peaxomz

mean .eeaeeez
meddmoopm soa
upompeoo spa: mmo

mfloa nmomfl .Lmaoom

mama .maonaa

mama .eeem

demonpmxomso

.m> mmmfippm Open: @ xoman

Lo anew Esme .m> anew unwaa

ASmOpsflxomz Lomv "pancammad maoa ma sofiumsfi
mmo m>apfimom uafimomad exp m>onm Amv Ga m< .u

mosmpmmmam pmaflmmm
UmsHMLp cons no .O>HpHmOQ
we: seams pea aduamaan

.m>fimao>m on» can: no pom: ma usmpm wmflgazm
.Umpsfimasd one; moasmmq m to .soamme Hmow
mesmfid doom npon smsz 62p Spa: msossapsoo mam dam
mac meapamoa easemaam smaaa mean: was Hana mesa
One .mmmaam dopsmflass .m>

ASmOpcfixomz pomv mmo oz Umpzmfia so open: .m> somam .m

HmOHpsmda mam

mode I cam + wsflzoaaom meson

62m .xon moagw so mumauw

2H .soflwop Hmom exp do “comm

CH Om to poop m heads as

ASmOQSHEOmz pmsammav SH hood Opes: no xomap m mam
mmo m>HpHmom mmdo maco u Opens .m> xomam .¢

soapmsfieromHQ anamw>

 

meoumwfipmo>sH

wcfldsflm Adeno: when; pgmoxo m>Hmeona memHv
:oHpmSpam msasmmma
no maze

 

 

.mpmm CH Ammov pommmm Hmmpo>mm wsassmmaem>o mo mammams¢an.m mqm¢a

 

pews

mpmdmg .mam: m haso AmpmsfieawpmdnHv

- Ame pmoma .:0pamao mmo meeeemez so eaeeeem
Home .Hamm

memesmz .mafinmxooam AQmOpsfixomz

mama .maonaa tome mmo eeapamoa

:oHmeHpo

nofipflmflsvom gm“: dew:

enema .eoaamflo Aemoeqaaeae some mmo oz

Asmpmm assess ompma
swan sesame mmama
mHHmSmH> mm; @009

mo muam use peso
-aeaamam eon mmov
acme .aeax e neaeeem

mac psm>maon
.maso pom p59 .qua esp msflon
menu go soapomafid Spa: .psm>m
uaophfl mama mam spon to .mmSo
mwmaumepxo .mOSO onmanmapsH .m

 

maddoooam
sofipooneoo spas
mmo doom mmmfi moapmpaao
. . mono HmSmH>
Lommoam a moamse vow 30H com: HH< mmmaumppxo a mmmaumnpsfl w:H©:Ho
Homa .nopfim @ nomH AsmOpsflxom: pmsammmv use .psmOQSOOL a psm>oamp mOSO
momfi .Honmmo mmo m>flpamom heme u weasemma mmmaue HmOHth .<
soapmsaaasomam Hmfipmmm
weepmmfipmm>sH wsfldqflm Admpo: mews: ugmoxm O>Hpsmosfl mmpmav

soapmSpHm wsasumma
mo maze

 

 

mama ea Ammov eooeem Hemeeeem amazemeaeeeo no weasemea ea Ae.eoov--.m mamae

10

I. Visual Discriminations

The first half of table 2 (pp. 8) presents the three
procedural variants used with visual discriminations. These
are as follows:

A. Trace conditioning paradigm in which the discrim-
inanda are not contiguous with the goal area;

B. Simultaneous conditioning paradigm in which the
discriminanda are maximally discriminable at the outset of
training;

C. Simultaneous conditioning paradigm with stimuli
that are minimally discriminable (most generalization) at
the outset of training.

For explanatory purposes condition B will be handled
first.f In this procedure the relevant stimuli fill the
whole alley and the goal area. Thus during reversal learn-
ing the competing response (Re) can be directly conditioned
to the cue that was originally positive (rewarded during
initial acquisition). The RO then generalizes to the same
cue at the choice point, and reversal learning is accom-
plished. In regard to overlearning reversal experiments,
the important factor is the extent to which Rc generalizes
to the new positive stimuli. If generalization is minimal,
that is, if the stimuli are maximally discriminable at the
outset, then overtraining is of no benefit and no ORE will
occur. This is the general result obtained. Since black
and white stimuli are distinctive for rats the Re to the

new negative cue during reversal is no more discrete after

11

overtraining than at the end of acquisition.

In condition C the same paradigm is used except that
minimally discriminable stimuli are employed (e.g. dark vs.
light grey; or checks vs. stripes). Thus when the acquisitin
criterion is attained the discrimination may not have been
perfected. Overtraining serves to perfect the discrimination,
narrowing the positive stimulus generalization gradient.

Thus Be in the overtrained group does not generalize to the
new positive cue as much as in the control group. Reversal
learning for the overtrained subjects is thereby facilitated
resulting in an ORE.

In condition A, a black-white discrimintion is used
with a trace conditioning paradigm. In this case the cues
are attached only to one-way valve doors that are located
a foot or so from identical neutral grey goal boxes. During
reversal learning, the Re is elicited in direct contiguity
with irrelevant stimulation. The one relevant cue is the
trace of the previously positive cue from the doors. Since
traces of stimuli undergo more generalization than prevail-
ing stimuli, the situation is analagous to simultaneous
conditioning with difficult discriminandum (c). Overtraining
produces a finer discrimination between the traces, per-
mitting the conditioning of Re to the new negative cue. Thus
ORE occurs.

II. Spatial Discriminations

In this group of experiments the procedural variations

far exceed those used in visual discriminations with the

results directly related to these variants. The second half

12

of Table 2 (pp. 9) presents a summary of these procedures.
It can be seen that the procedures fall into two categories:

A. procedures with all cues relevant and redundant

B. procedures with some cues irrelevant

In the procedures where irrelevant cues were present
or introduced (Table 2, IIB), both a positive and negative f
ORE have been obtained. It must be pointed out here that for
Mackintosh's theory of ORE, the presence of irrelevant
stimuli is a sufficient condition for the manifestation of
an ORE. He postulatestflmfliovertraining serves the function
of "switching tn! thnerblevant stimulus dimension and
"switching out" the irrelevant ones. For Mackintosh however
no further analysis can or should be made.

In order to use elicitation theory one must submit to
scrutiny all the procedural details of these studies.
Pubols (1956) and Brookshire, Warren and Ball (1961) both
obtained a positive ORE. Pubols used the same apparatus
to study spatial discriminations as he did to study visual
discriminations. In his visual discrimination study,(Pubols,
1956),the discriminanda were black and white one way doors
set half way down the grey arms of the y-maze. It was
explained earlier that using this paradigm the rat can
associate only the stimulus traces of the discriminandum to
the reward contingencies. In his spatial discrimination
study (Pubols, 1956), these doors were the only irrelevant
stimuli. Thus the likelihood of the rat incorrectly using

irrelevant trace cues instead of relevant and salient

13

extra-maze cues which are present both at the choice point
and goal area, is small. This makes Pubols' experiment
most similar to Spatial discrimination studies classified
under (A) in which ORE occurs.

The Brookshire, Warren and Ball (1961) experiment is
more complicated. They used a cross-maze in which the
animal began equally often in each grey stem. The irrele-
vant stimuli consisted of one arm being white and the other
black. The direction of the turn was supposed to be the
main relevant cue. However, using this arrangement, the
rat could have 1e3nned:a conditional discrimination during
the 120 overtraining trials. That is, it learned to
approach black when leaving start box #1 and to approach
white after start box #2. Thus, an overtrained § during
reversal could be using a compound stimulus including the
trace of the start box cues plus the black or white of the
arm. This situation is most similar to visual discrimi-
nation classified under (A) where a positive ORE is ex-
pected, since the stimuli from the start box were only
perservative traces when reinforcement occurred. It must
be pointed out that this analysis is only speculative and
is not; supported by direct evidence. However, it does
suggest that a recategOrization is reasonable.

The only study classified under (R) of Spatial
discrimination which obtained a negative ORE was Clayton
(1963b). In this study Clayton used horizontal vs. ver-
tical striped panels placed on the floor of the choice

point and on the one-way doors preceding the goal box as

14

irrelevant stimuli. These were randomly changed from trial
to trial. The direction of the turn was defined as the
relevant cue. There was no attempt to scramble extra-maze
stimuli nor to control for unspecified intra-maze cues
(i.e. differential olfactory, floor texture cues etc.).
Thus an exact analysis of the stimulus conditions in this

study is very difficult to make.

STATEMENT OF PROBLEM FOR EXPERIMENTS I, II, III

Due to the absence of a conclusive study using irrele-
vant stimuli in a Spatial discrimination three studies were
undertaken. In the studies just reviewed stimuli were
introduced and then an attempt was made to make them irrel-
evant. In the present experiments the extra-maze and intra-
maze stimuli were already present in a spatial discrimina-
tion, and all of these stimuli were made completely irrel-
evant except directlion of turn, right vs. left. In this
way, one can specify what stimuli the animal has to be
using. This should greatly facilitate the analysis of the
results when using the elicitation position. In experiments
I, II, and III, both the size of the incentive and the
length of the stem were manipulated in an attempt to ascer-
tain the effect of overtraining on the subsequent reversal

of a pure Spatial task.

15

EXPERIMENT I

INTRODUCTION

 

This study used a small incentive and a relatively

long stem.

METHOD

Subjects,

Ten male albino rats from Spartan Research Animals,
approximately 200 days old at the beginning of the eXperi-
ment, were used as S's. They had been used previously in
a shock study being run for one day at an age of 90-100

days.

Apparatus

 

The apparatus as shown in figure 1 was a T-maze with
movable stem and start box, constructed of plywood floor and
1.90m. pine sides, and was covered with hinged hardware
cloth. It consisted of the following parts: starting box,
runway, two arms and a choice point. The start box was
34.4cm. long. The runway was 18.8cm. long and the arms 35.6cm.
long. All parts were 7.6cm. wide and 7.6cm. high. The

choice point was 7.6cm. square and 30.0cm. high.

16

 

rat cage
assembly

 

 

rat cage
assemny

 

SIDE II

 

 

I'"""I
I l
I I
I I
I I
I I
I I
I l
I I
I I
I I
l I
L---I
I'""'l
I I
I I
I I
I I
_ I -__.l
arm....... .
I 3 ft. , food %a
. 7IdlSh . 1. *
0110108.... 00.000.00.000. *0;

 

 

 

 

 

point F—'" "
run‘Nayoooooooooooooooo vertical

sliding
vertical sliding...... doors

door

start box.............

 

 

 

SIDE I

nnnnnn

diffused overhead
florescent light

Fig. 1 A diagramatic representation of the
running room and T-maze used in
the experiments.

 

riatauzz

HH>S

17

Five vertical sliding doors, manually operated, were
employed in the maze. One separated the start box from the
runway and four were on the four sides of the choice point.
The doors on the choice point were necessary since a non-
correction procedure was used and the procedure called for
alternating the start box and runway to opposing sides.

The entire maze was painted flat black and the tOp
22.9cm. of the choice point was covered with flat black
cardboard. Thus, the only illumination in the choice point
came from the Open sliding doors facing the arms. These
doors were open to 5.9cm. from the maze floor.

The extra-maze stimuli in the running room consisted
of a white wall to the right of the maze perpendicular to
and butting the right arm and a rat cage assembly perpen-
dicular to and .9144 meters from the left arm. The over-
head illumination was diffused forescent light to the
left of and behind the choice point. Figure 1 diagrams

the floor plan and apparatus.

Procedure

 

Preliminapy training

The S's were handled for 10 days prior to any eXper-
ience with the maze. The first 10 days of handling were in
the following order. The rats were placed on 24 hour food
deprivation (1 day) prior to the first day of handling.
The first five days of handling consisted of holding rats,

transferring them from hand to hand and placing them back

18

in their individual living cages with three large food
pellets (190mg.) on the floor of the cage. The next five
days consisted of continued removing and replacing the rats
in individual feeding cages baited with two 190mg. Noyes
pellets. This was continued until the rat reached the
double criterion of being placid when held (i.e. no struggle
against E's hand) and eating the pellets within two
seconds after replacement in the feeding cage. The next
four days the rats were allowed free access to the entire
T-maze for five minutes per day. Both arms of the maze
were baited with adlibitum food (45mg. Noyes pellets). By
the fourth day all rats were eating the pellets in both
arms for an equal amount of time. On the fifteenth day
discrimination training was started.
Discrimination training

The reward size was two 45mg. Noyes pellets placed in
the food cup at the end of the arm. All rats were trained
to discriminate the kinesthetic stimuli of turning, half
trained to make a right turn and half a left turn. This
was accomplished by alternating the start box 180 degrees
to opposing sides in a quasi-random schedule. The response
designation dictated the placement of the reward. Thus
a rat trained to make a right turn would have to enter the
right arm and approach a white wall when started on side I,
(see figure 1). However, he would have to go to the left
arm and approach the rat cage assembly when started on

side II. The arms were kept in a constant position

19

throughout the experiment. By this method the only relevant
cues associated with reward were the proprioceptive stimuli
generated by turning.

All rats were given twelve trials per day, with an
inter-trialiinterval of fifteen seconds. The criterion for
acquisition was 83 percent correct or more in one day (10
correct responses in 12 trials). Upon completion of the
discrimination task, the animals were assigned to groups so
that the mean trials to criterion of the groups were as equal
as possible. That is, if S's score was higher than the
cumulative means of the two groups then it was placed in the
group with the lower mean. Conversely, if an S's score was
lower than the cumulative means then it was placed in the
group with the higher mean. When S reached criterion it
was assigned to a group and the appropriate stage of training
(i.e. overtraining or reversal) was begun the following day.
The experimental group received 96 trials more.r training on
the original discrimination and then given reversal training.

The fixed number of trials of overtraining was obtained
by taking 200% of the median trials to criterion (Md.=48 trials).
The control group was reversed immediately.

Reversal training

 

During this training the original negative cues were
now associated with reward. The reward again was two 45mg.
Noyes pellets. All rats were given fifteen trials per day
on the reversal task and were run to a criterion of 80 percent

or more correct in one day ( 12 or more correct in 15 trials).

RESULTS

Table 3 presents the results for original learning and
reversal learning. The measure of performance used for all
stages of this experiment was the mean number of trials to

a criterion including criterion trials.

TABLE 3.--Mean trials to criterion for original acquisition
and reversal of the experimental and control

groups.

 

 

STAGE OF EXPERIMENT STANDARD
Group N MEAN DEVIATION

 

ORIGINAL ACQUISITION
Experimental (overtrained) 5 62.6 13.9
Control (no overtraining) 5 76.8 19.6
t=1.73; d.f.=8; n.s.
REVERSAL ACQUISITION
Experimental 5 81.0 7.38
Control 5 51.0 15.29
t=3.54; d.f.=8; p.<.OO5

hn

Opiginal Acquisition

The mean trials to criterion on the original learning
for the experimental (overtrained) group was 62.2 and for
the control group it was 76.6 trials. There is no signifi-

cantLdifference between the groups (2:1.73; §,£.=8; p<.1)-
20

21

However, the groups were not closely matched with regard to
learning rate.

Position preference was assessed from the first 2 trials
of original acquisition. If an S produced two incorrect
responses in the first two trials it was classified as
having a strong negative preference, i.e.,a strong preference
opposite to the designated correct response. If an S pro—
duced two correct reSponses it was considered to have a
strong positive preference. An S with a weak negative pre-
ference would first produce an incorrect response and then
a correct response. A weak positive preference Was a correct-
then-incorrect reSponse order.

Table 4 presents the number of SS in the experimental
and control groups exhibiting a specific position prefer-
ence. From this table it can be seen that the groups were

TABLE 4.--Number of SS in each group exhibiting a specific
position preference.

 

 

 

Strength of Direction of Preference
Preference Group Group Total
Positive Negative
Exp. Control Exp. Control
Strong 0 1 2 O 3
Weak 1 1 2 4 8
Total 1 2 4 4 11

 

approximately counterbalanced for direction of preference
with the majority of SS having weak preference. There were

two SS in the experimental group with a strong negative pre-

22

ference and one S in the control group with a strong positive
preference. One would predict that a strong negative preference
would lead to slow acquisition ' ' i ' of the original
response and faster reversal. A strong positive preference
would produce the Opposite effect. The individual data for

the acquisition and reversal of the three Ss with strong
preferences are opposite to the predicted result. Thus,
position preference can not have systematically confounded

the obtained result.

Reversal Acquisition

 

For reversal, the overtrained group reversed signifi-
cantly more slowly then the control group (t=3.535; d.f.=8;
p<:.005). The mean trials to the reversal criterion for
the overtrained group was 81 and for the control group it
was 51. Due to the fact that the experimental and control
groups were not well matched on trials to criterion, an
analysis of covariance was also performed. Table 5 presents
the summary of the analysis using trials to criterion for .
acquisition and reversal.

TABLE 5.--Summary of the analysis of covariance with trials
to criterion as the measure.

 

 

 

Source of Sum of Degrees Mean

Variance Squares of Freedom Squares F V p

Amt. of training 2008.419 1 2008.419 10.260 p<.025
Error 1370.220 7 195.743

Total 3378.439 8

 

From Table 5 it can be seen that the effect of overtraining

23

was still significant (F.=10.260; df.=1/7; p<.025). Thus

a significant negative or reverse ORE was obtained.

DISCUSSION

Experiment I showed that overtraining retards subse-
quent reversal learning of a pure spatial (i.e.,kinesthe-
tically controlled) reSponse. The results plus observations
of the actual behaviors of the rats while they were running
in the maze led to the following explanations and predictions.

For the overtraining group it is postulated that the
stimuli controlling the proprioceptively controlled turning
response had backchained down the stem away from the choice
point and goal box. This is partially supported by the
observation that, late in training, all overtrained rats
would typically start turning when the start box door was
raised. They would also orient in the start box so as to
facilitate the early turning. Thus, if an animal was to
make a right turn, its nose would be pointed to the left
side of the start box door with its hind quarters pointed to
the right rear of the start box. This facilitates making
a centrifugal swing to the right. Thus it may be said that
the initiating stimuli of the chain could have backchained
to the start box. Since these stimuli were not temporally
and spatially proximal to the site of non-reinforcement
during reversal they could not readily be associated with

the frustration and withdrawal reSponse necessary for the

learning of reversal. The stimuli for the immediate rever-

21;,

25

sal group, due to the lack of overtraining, had not back-

chained very far from the potential site of frustrations,

Thus, these stimuli could be more easilty associated with the
frustration-elicited withdrawal response. This presumably,
resulted in faster learning for the control group.

The next two studies were designed to investigate the
effect of the length of the runway and reward size upon the

reversal of a kinesthetically controlled turning response.

EXPERIMENT II

INTRODUCTION

 

In EXperiment I it was postulated that overtraining
retarded reversal because it lengthened the S-R chain
associated with reinforcement making the initiating stimuli
distal from frustration during reversal. In this experi-
ment the runway was completely removed. The start box was
shortened and attached directly to the choice point. It
was reasoned that this would effectively limit the length
of any chain formed during overtraining, decreasing the
effect of overtraining seen in Experiment I. Large reward

was used in this and the next experiment.

METHOD

Subjects
Twelve male albino rats from Spartan Research Animals,
approximately 200 days old, were used as S's. All rats had

prior experience in a shock study at the age of 90-100 days.

gpparatus

 

The apparatus was the T-maze described in Experiment I.
The following alterations were made on the maze for this
study. The runway was removed, and the length of the start
box was shortened to 18.8cm. The start box was placed at
the bifurcation and the door of the choice point was used as

the start door.

Procedure

 

Preliminaryrtraining

 

This was identical to that used in Experiment I.

Discrimination training

 

The training procedure was identical to that used in
EXperiment I.

Reversal training

 

The reversal training was continued until all rats
attained a criterion of at least 80% correct in a block of

15 trials given in one day.

27

Table 6 presents the mean trials to reach learning

criterion for original learning and reversal.

TABLE 6.--Hean trials to Criterion for acquisition and
reversal of experimental and control groups.

 

 

STAGE OF EX??? IEET

 

STANDARD
Group N NEAR DEVIATIOL?
ORIGINAL ACQUISITION
Experimental (overtrained) 6 58.5 19.576
Control (no overtraining) 6 50.0 17.5

t=.723; d.f.=10; n.s.

RE‘ERSAL ACQUISITIOI

Experimental 6 60.0 8.660
Control 6 45.0 12.247

 

Original-learning

 

A t-test performed on the mean trials to criterion of
original learning yielded a of .733 (dleo) which was not
siomificant Although not sibmificant, the Spearman rank
order correlation between original acquisition and reversal
scores was .73 for the experimental group and .33 for the
control group, necessitating a need for an Analysis of Co-

variance.

N
I)

“x

Position preference was assessed as in Experiment I
by the score on the first two trials of original learning.
Table 7 presents the number of S exhibiting each type of

preference.

TABLE 7.--Ku:ber of S's in each group <Xhibiting a specific
position preference.

 

 

 

Strength of Direction of Preference
Preference Total
Group Group
Positive hegative
Exp. Control Exp. Control

Strong 2 2 0 0 4
Weak 1 0 3 4 8
Total 3 2 3 4

 

It is evident that the groups are clearly equated as to
position preferences.

Reversal learning

 

The overtrained group reversed in a mean of 60 trials.
The control group took 45 trials. This difference is sig-
nificant at the .05 level (t=2.236) (df=10). However, the
Analysis of Covariance (see table 8) on the trials to re-
versal criterion suggests that overtraining did not signifi—
cantly retard reversal.

TABLE 8.--Summary of the Analysis of Covariance

 

 

Source of Sum of Degrees of Mean
Variance Squares Freedom Squares F p

 

Amount of
Training 459.433 1 459.433 3.957 n.s.

Error 1044.847 9 116.094
Total‘“ . 1504.280 10

 

The fact that no ORE was found in Experiment II
suggests that either a relatively long stem in the T-maze
or a small reward, or both, accounts for the negative ORE
in EXperiment I. Experiment III was done in an attempt to

identify the relative role of these two variables.

30

EXFE iZI31 3.1T III

*fivgl

 

This .xper1:ne:1t differed in one way from I‘xnerinent I

(

and in one way from Ixnerinelt II. Experiment III differed
fron Experiment I in that a larse reward was used. This
should serve mainly to increase the elicitation value of
non—reinforcement (i.e. frus tration-withdraral response);
this in turn should facilitate the acquisition of the original
response, especially for Ss trained against preference
(Denny and Dunham, 1951). More importantly the increased
size of reward should facilitate the acquisition of the
reversal response, attenuating a negative 01E. Thus a
lesser tenden y toward a negative 03E should be found in
Ex mp rinent III than in Expe rinent I.

Experiment III differed fron Experiment II in that
a long start box plus sten were used. This should increase
the effect of backchaining by increasing the potential leng- gth
of the chain. Thus a greater tendency toward a negative

0E1 should be found in Experiment III than in Experiment II.

31

METHOD

Sub‘ect.

1,

Twelve male albino rats from Spartan Research Animals,
approximately 200 days old were the subjects. No subjects
had received prior shock exposure.

Apparatus

 

The apparatus was the T-maze used in Experiment I. It
had a 34.4cm. start box and a 18.8cm. runway.

Procedure

 

Preliminary trairing

 

The handling and feeding procedures were identical to
those used in all previous experiments.

Discrimination training
L.)

 

This procedure was identical to the one used in Exper-
iment I and II. All rats were trained on a turning discri-
mination, reward size being two 190mg. Reyes pellets. After
an acquisition criterion of at least 83 percent correct in
twelve trials is attained, the S's were assigned so as to
equate the means of the experimental and control groups on
trials to criterion. The eXperimental group received
200 percent overtraining trials and then were reversed. The
control group was reversed immediately after reaching cri-
terion.

Reversal training

 

All S's were run to a reversal criterion of at least

32

33
80 percent correct in fifteen trials presented in one day.

This is identical to Experiments I and II.

Original acquisition

 

Table 9 pre.ents the mean trials to criterion for the
original learning. The experimental group acquired the
original response in a mean of 40 trials.

TAELE 9.--The mean trials to a criterion of 80 percent correct

in any one day of original learnirg for
experimental and control groups.

 

 

 

GROUP N MEAN STAEDAED
DEVIATICT

EXPEEIHEUTAL 6 40 23.66

CONTROL 6 38 23.41

t=0.6518; d.f.=10; n.s.

 

The control group obtained a mean of 38 trials. This
difference is not significant yielding a t of .6518. It
must be pointed out that the variance for both groups was
higher then that obtained for EXperiments I and II. The
trials to criterion ranges for both groups from 12 (i.e. the
first day of acquisition) to 84 trials. This large a range
was never obtained before.

Table 10 presents the number of subjects per group

demonstrating a positive or negative preference. It can be

seen that the experimental and control group were approxi-
mately equated in the direction of preferences with the

majority of subjects demonstratin" weak preference.

6.)

34

35

TABLE 10.--Eumber of subjects demonstrating preference.

 

 

 

Degree of Direction of Preference
Preference Group; Group
F sitive Iegative
Exp. Control EXp. Control Total
Strong 2 i 1 O 4
Weak 1 3 2 2 8
Total 3 4 3 2 12

 

The three g's in the experimental group and the one s in the
control group did not demonstrate the appropriate change in
learning rate from original acquisition to reversal as would
be predicted if position preference mediated the experimental
effect. Thus preference could not account for the difference
between the groups found in reversal.

T3 -°.N
neversal learning

 

Due to the large variance foun with the trials to
criterion measure in both original learning and reversal,
other measures of performance were tried. The measure that
was used for comparisons was trials to a sliding criter-
ion of five correct out of ten trials. The criterion for
using this measure was that it had the lowest variance of
all measures tried. In order to decrease further the
amount of error variance a two by two Anovar was performed.
rhe subjects were placed into one of two categories, fast
or slow learners, based upon their original acquisition
scores. This was done by dividing the range (72 trials) in

half obtaining a demarcation line of 36 trials. Those S's

"N

\41)

that learned the original discrilithion in less than 36 trials
were considered fast learners. Those that took more than

36 trials were slow learners. This method discriminated

well since no S learned in 36 trials; the fast learners

ranged fron 12 to 24 trials and the slow learners from 48

to 120 trials to criterion.

Thus the Anovar compared the effect of two variables
upon the reversal scores: original learning rate (fast vs.
slow) an d the amount of training (overtraining vs. no over-
training . It should be pointed out that all reversal
measures tried showed a trend toward a negative ORE.

Table 11 presents the summary of the Analysis of
Variance.

TAB E 11. --Sun nmary of the Analyr sis of Varianc 0e using trials

to a reversal criterionzn5 correct out of
10 trials.

 

 

 

Source of Sum of Degrees of Mean

Variance Squares Freedom Squares F P
Learning Fiate 1,386.750 1 1 ,386. 750 b? 669 p< .001
Amt of Training 234.083 1 234. 083 7.202 p<:.05
Interaction 154.084 1 154. 084 4.741 14:.1
Error 260.000 8 32.500

Total 2,034,917 11

 

It can be seen that both learning rate and amount of train-

ing are significant with F-ratios respectively 42.669 (p¢:.001)
and 7.202 (I<:.05). The interaction was not significant at

the .05 level but was S1ignificant at the ten percert level.

The overall mean trials to criterion (five correct out of

ten trials) for the overtrained group was 32.22 and for

the control group it was 23.3 trials. Thus the results of

or?
.21

this analysis demonstrate an overall negative ORE.
Table 12 presents the means and variance. It can be

seen from this table that the most pronounced effect of over-

training is with slow learners. For the fast learners the

200% group is only 1.74 trials slower than the OT group.

For the slow learners this difference is 16.01 trials. In
the former case this is a 9S retardation in the latter a
8% retardation.

TABLE 12.--hean and variance for the four groups used in
the Anovar summarized in Table 11.

 

 

 

 

 

ACQUISITIOI RATE AMOUhT OF OVERTBAINIHG
2005 egp. group Qfi control group
Standard Standard
N Mean Deviation K Kean Deviation
FAST 3 18.00 2.828 3 16.33 2.896

Slow 3 46.67 4.120 3 30.66 7.320

 

DISCUSSIOX

The interaction between learning rate and amount of
overtraining makes sense with the analysis of the negative
ORE obtained in Experiments I and II. It will be recalled
that overtraining was postulated to elongate the behavioral
chain that results from a kinesthetically controlled response.
This occurs through the addition of the individual links of
the chain which follows the laws of classical conditioning.
It is generally accepted that there is a point where a
further increase in the number of trials has little dis-
cernible effect on learning. I am saying that with fast
learners the aysmptote of learning 's reached earlier for
each link of the chain, allowing a rapid elongation of the
entire chain. Thus it is possible that the fast learners

in the immediate reversal and overtraining reversal groups

|_J

had engthened the chain as far as possible at the end of
original learning. Thus, overtraining would not separate
the fast learners as much as the slow learners since the

former would have less room for elongating the chain than

‘40
(‘0

EXPERIMENTS I, II, & III
COECLUSIOS

From the point of view of hack'ntosh, the presence of
many salient irrelevant cues and a concomitant difficult
discrimination should result in the occurrence of an CHE.
However, as it has just been shown, the Opposite effect can
occur.

The procedure used in EXperiments I, II, and III was

designed in order to Specify exactly what stimuli and re-

.L

’J

eponses were being used by the subject. This makes the

k

“5
(D

sults amiable to analysis using Elicitation Theory. This
procedure places all of the stimu us control on the cues
which result in making the appropriate turn. These cues
were present at the choice point or even before (kinesthesis
attendent upon the centrifugal swing in the stem of the
T-maze). Thus, during reversal training the competing re-
sponses elicited by non-reinforcement (fr stration-withdrawal)
did not occur in close temporal contiguity to the new neg-
ative cue. Only a perservative trace of kinesthesis was
present when the competing responses are elicited. Thus the
effect of overtraining was as follows:

a) through backchaining the critical kinesthetic cue
:as pushed back down the stem further away in time and
Space from the locus of the frustration elicitation of Rc

during reversal training, resulting in a reverse ORE.

e l

(1‘
bd
(D
H)
C1-
I

b) with more trials the cerservative traces of

K
a;

right turn may have become more discriminable, as de crib-

C)

earlier, decreas'ng but not overpowering the first
effect., This might yield a less substantial negative 33
when large reward was used.

The above analysis receives support from results ob-
tained by hackintosh (1965) which show that when all cues
are relevant, kinesthesis (direction of turn) acquires an
increasing proportion of stimulus control as the number of

trials is increased.

SKIES KEITS IV a V
IhTEODUCTION

In order to round out the picture of 023 in spatial
discriminations EXperinents IV nd V were performed. These
experiments are complementary to,but quite different from,
the first three experiments. In Experiments IV and V all
cues were relevant and redundant and the T-maze used had
a shortene stem to minimize backchaining. Together, these
two studies investigate the effect of the strin ency of the
acquisition criterion, the amount of training, and the size

of reward upon reversal learning.

hi

71‘“ f‘:’\'Y-‘|T'\ " ,' 'I'fil 'm I‘J
a - m 1 ‘ ‘ -' '

“JFLJv Al- ‘ -‘4‘T .L

LIES ‘HGD

Sub‘ects

._.__....H-'_____
Twelve male albino rats from Spartan Research Animals,
approximately 200 days old at the beginning of the experi-

ment, were used as g's. They had been used previously in

a shock study being run for one day at an age of 90-100 days.

Apparatus

 

The apparatus was the T-maze described in Experiment I.
The following alterations were made in the maze for this
studI. The runway was removed, and the length of the start
box was shortened to 18.8cm. The start box was placed at
the bifurcation and the door at the choice point was used

as the start door.

Fri

 

Preliminary training

This procedure was precisely the same as in Experiment I.

. . v‘f . 1V . ' L r\ . \l‘ . ‘I
DiscrimiratIOh training

 

The start box remained for the entire experiment on
Side I. half of the Q's were raining to go to the left arm
and half to the right arm 0 the maze. They were given

-n
~

2—453g. noyes pellets as reinforcement. When the rats

1;.2

Ls

(Hm-'3’ ‘4’}

reached a criterion of up” or more correct in I; trials given
each day, they were matched on trials to criterion and ran-
domly assigned to two groups. The experimental group re-
ceived 51 overtraining trials on the original response and
then reversed. The amount of overtraining was determined in
the same way as EXperinents II and III. The control group
was reversed immediately.

Reversal trairin"r
Reversal training was the same as Experiment I. All
Q's were given 15 trials per day until they reached a rever—

sal criterion of BOfl or better in one day of train'ng.

In this study two easures were used to assess rer-
fornance. ”hese were trials to a performance criterion of

BOfl correct on ary one day, including criterion trials and

trials to the last error, excluding criterion trials.

Table 13 and 1M :resent a summary of the results for acq-
uis iticn aid reversal 'espectiv ely in both groups.
TAEUE 13.--Summary of the acquisition data for t o reasures

and both groups.

 

 

 

 

 

S tandard
Group H Kean Deviation
“"neisnunital
0
Trials to Criterion 6 2a.? 1.501
rials to Last Error 6 15.0 5.7?3
Control
“"-1 t fr't * ” “6 l W?”
_r1a s o s-1,eri01 o z .7 L.Y-o
Trials to Last Error 6 13.3 0.9e1
Crigiral Acquisition
Table 13 preserits the sum;ary of results for the exp-
erimental and control groups. The groups did not differ

significantly on both measures at the outset of the exp—
eriment. Trials to criterion and trials to the last error
produced t—test results es ual to 0.951 aM 0.332 respectively

(d.f.=10; n.s.)

Lug,

kn

Table 1% presents thi su:m1ary of results of the rever-
sal sta“e of this "oerinent. There was no difference be-
tween the exceri ~ (ental and control groufs whei trials to

f the reversal (ata for three measures ad

' rm? fl {1 x r
TlsiE iu.-—sum a-) o
for ooth groups.

 

 

Standard
Group h Kean Deviation

 

ESL? 0T1. 18'7““ '11

Q\
i.)
O
O
(3
(3

n J... ‘ o _ ‘ n ‘
Trials to Cfltflflon

Trials to Last Error C 19.5 5.530
C03t1‘07

Trials to Criterion 6 30.0 0

Trials to Last Error 6 8.7 n.395

 

criterion was compared. Since all reacned criterion in 30
trials there w.s LO variance. It was cone ded that trials
to criterion was too gross a measure to detect differences
when lecrhir< was so rapid. Thus, trials to the last error

was used. Kean trials to the last error was significast

(t=3. M35; dfziO; p <.OO5) with the experimental grOL.

averaging 19.5 and the cortrol group averagih“ 8.7. It rust

p—J

order correlation be-

0
O

‘- , oihted out the the Spearman New
tween cflcru sitio and reversal scores on trials to the last

error was —.0M

cortrol group. This makes it highly unlitely that the ob-
of L

1‘.)

difference was due to di

'fl.
:

ta

”I (‘3
.. J

A . D O / r” ‘, 7 “A . ~-\ —
erencv Oi two roups. ihus tee or r :‘Ql.OL

WC EVQTRWQ EOPO QT?OPC 0V9? 40?? trials than

control ”r 'e “ ~ ,
. a ”oug. so scque”tly a reverse 0V7 * " ‘ '
-l its. lei was ootained.

k)

EXPE:;II1EL'T v

To clarify the results obtained in Experiment IV, EXp-
eriH ment V was performed. here reward size and level of
acquisition were manipulated to determine their effect upon

the reversal of a place response.

1 AETEiOD

Subjects

The g's were 2M male albino rats obtained from Seartan
Research Animals. All were 200 days old at the beginning
of the eXperimeht and were used previously in a shock study

at 90-100 days old.

Apparatus

 

The apparatus was the T-maze used in Exper i1 ert IV.

Preliminary trairing
The handling and habituation procedure was the same as
Experiment I.

DiscriflihatiOh trainihm

 

efore training the g's were assigned randomly into

4 groups in a 2X2 Factorial design. The four were as follows:

47

A , ~ .- q 2 I r‘ "' I ’1 " "‘ 1 ’2‘
cVeileainih5-snali .1 arm, 0vezl-i hin5-la-5e I w 3 Iumfiul-

“ ~ r ' -. ‘ 1 .7 , “u"
ate revrrsal ll reward, and Irnediaic reve rsal-lar5e retard.

All rats were trained on a place discrimination as in Experi-

nt IV with all e: :tra and ir tra ease and kinesthetic cues
relevant. The animals in each group were haphazardly ass i5hed
to the respoz1s e desi5nation of ri 5ht or lefta m“n correct with
the provision that half of each group was assigned to each
side.

The overlearning groups were run to a criterion of at

least 80$ correct during a daily block of 12 trials. They
were then given 1005 (i.e. 24 trials) overtraining. Reversd.

riterio on nof at least

0

groups were trained to an acquisition

LY

5 correct out of a sliding block of 6 tr

:3.

als. here acqui-
sition was continued until the Q's reached criterion. he
rext day they were reversed. This manipulation was used to
sarplc the effect of very low levels of acquisition and
overtrainin5 upon reversal. The small reward group rece-
ived 2-45m5. Ecyes pellets throughout all sta.5‘ of the
experiaert. The large reward was 2-19Qz5. pellets through-
out. The total int ake for each ur1*’l was balanced by
manipulating the o laiitity of he :e cag efood, so all animals

”’4.

received an amount of food.

CD
pa
5:;

{-1

During reversal all Q's were run to a criterion of at
least 80: correct in a block of 15 trials per day. The
Q's received the same quantity of reinforce ..... Nit they did

1

lEtULTS

H]

L

"x"

(D

In this study t1 main measure of performance used
was trials to the last error, excluding criterion trials.
The subjects learned the original task and the reversal of
the task so fast as to render trials to criterion a gross,
nondifferentiating measure (see Table 15a). Acquisition
data for the immediate reversal and overtraining groups
were not compared due to the difference in acquisition
criterion for the two groups. Table 15 a & b presents

reversal data for all four groups.

TABLE 15a.--Trials to Criterion during reversal for all
four groups.

 

 

 

GROUPS TRIALS TO CRITERION
Standard
N Mean Deviation t
LARGE BEWA21
Experimental 5 18 6
1.?llflqs
Control 6 17.5 5.590
SHALL REWARD
Experimental 5 33.0 6
1.260ns
Control 6 30.0 0

 

 

 

 

“V“ f“, ' . 1‘ ’~ Pr‘V'L ’1 *n r1 1"": ’) " "1’3!“ fps
Tazlflz 1;L).--_LI’J-C1--C to Cu“ . a ..‘..L PO- ‘Lv.- ll-“ 1 C 19.. no.1 cl”
all lou; Droufs.
T7 "TT'I‘C‘ m“) T .“ T ('3 mp T A C‘I’T‘I 17f'\'_1_f‘..‘!’)
.-- l.) p :- Jp-£‘-xLJ—U uh V’ J—J‘AU uh L“ ‘A V’-‘
C‘ t w r‘
standarl
‘1- q., Fifi n a J" o _ t
Le gain uGVlablOfl
T A v1 1" H‘ 7‘1—
ua vs ‘;..u; ‘ D

L1 1
“4
:5
(D
’3
}_ o
5
c%
{D
F—J
\n
(‘0
O
‘ D
C
H -\)
{:-
L

O\

b.)
{441775 N

C)

N

F4) .
O
H
\_K

C‘“,‘.TT yrv'nr-

7
’ -. ' . " 4 ,.
o--A_.4—_ fi‘“..‘--~

O
‘ ‘- '_ ‘1 r ‘ ‘ ‘ r}
:_.l..T»01 l-;n31-tu.l

J-

U1
[\

(3
O
:33
(1'
w
0
IL--'
0 \
N
'3
O
O
K.)
O
,3
\J

 

Table 151 that a significant pOSitive 23 was obtlirod with
larva reward whereas a significant negative 032 was obtaired
uSing a small reward. In order to test the significancv of

the interaction between reward size and amount of training

-+

V ...4 r v' C O ' 1 0 o
a 222 “govar was performed. since JfllS analySis assumts

equal 0211 SiZLS one 8 was ranflomly deleted froa each of the

large and small reward is odiate reversal groups. Table 16

 

 

 

~~~~~~ / . -n". .I- .w .- n v. .11“. . — .~ ~ -~ V: ‘~,
lulll Ti.--Sup tr; satin ,9 the Analgsis of Varia cc .it:
trials to thc Last Error as a measure.
n ' - 0 ~' " w- w u m" " -\_ 3"
source of un of D Drugs cl Utah
. - - ,. r‘ . i ‘ 1:). -l.) A\ ‘1 .- (‘I 71
Variation squares grthqQ quere : p

 

1“. ,,- . .-J-
iluouut of
: . n
’Iiflll-1l:;J oiJOC) l
' - ' a o
EaC hituac Oi
r30 I"I'3 [“11 mod 1

T" l l

Lr Yam 7‘?
*¢.-LJ\_I.. 'aVC L'd—

[—8

Oil

muvsoOZ-n

A—J‘IJ-

14

C\

Total

0
.COO
1&41 2 11‘
r109 I) rial;
/\,.'v.l\ #-

.00; ns
rdff " (\‘r"
flat-3 . n
.VV/ l<.O‘-_~.1

 

3rom t.o Table (16) it is
reward and the reward by
are 2 graphically Iresehts the
size of rvuar as the pa'ametcr.
cZE MKCrcES slall r var

vvlq

aIparent that

training interaction was

‘Tj

Larg reward J-

8

id results in the eXpectod

u. 900-4151- t’ lee

interaction effect

of
significant.
w i th

ields ar

\
L

negative ‘“

51
a-a—d .

to Last Error

Trials

284
26‘
244
22—
20-
18-
1 6..
1 LL-
12%
10“

8d

6-

 

‘7
IfiInnwrn.--Graphic presentation of the interaction of
incentive size and amount of training
with trials to the last error as the

52

 

23

Number of Trials Overtraining

O-b

measure.

g.___.; Small Incentive

55-..-.. Large Incentive

"Go. 00 Main EffeCt
(training)

D.

..

 

E‘"""T'T.'?TT‘T"fI-’ “'2 '. *7
L4'.. 4.4. J—Alu$i&u .L

“Wfir‘vrr‘(‘|~”r\"

H—L—a‘v V‘JH-LV-‘

4- rw T

‘ v ‘ , (:1 . 17 r 1 1 . h- ‘L
an explanationc of the result of en :rire us i a J

stems from the assumption that the eff- ct of overtraining

L

a) whether mall or larg e reward was being used, the

approach response to the positive -timuli was beiig stren-

CO

thened by increased number of trials, this alone would

'Le exnected negative ORE.

‘-

:4
H
O
h‘
L.‘J
C.-

m
a

b) SilC e the acqui is'tion criterion for all studies w

low the total nUaeer of trials in original learning did not

" K . f.‘ f) 4" . v '. ‘ 1x ‘ (j . x , o . r- r 4- v1

exceed ;; trials with one a outain ré this and cLe rest
0 ‘ 'V' . P.“ 4" O . f" (N .

o; the s's lgaruino in 12 b0 1o trials. since the 10‘

0
<1
TI
,1
C+
‘fi
e
H.
'_.
F).
'r
F“
(4
O
C.
;__1
CL
9
\_
.5“
9
5
(D
St
6
Ii
".3
C
7c
T‘
4

_ Vod
.L.

9
O
H)
r!-
L?“
(D
'1
C
7
r

t.

C'-
i...“
Q
1")
H)
H:
C
O
(“1"
e3
C
C
“‘5
'7)
C 1"
{5‘
(D
L _)
J.
3
C?‘
1')"
Q
r
’ :5
O
{—1-

/ 1 .. U, r. -A - .
(Alsel 1902). In tneonr this deans more COLSlStCLt
eliCitatien of a yore vigorous withdrawal res Ionse to the
new nega ative cue during reversal. Several studies indicate

that the frml tratien effect increases with the number of

2.)
c...
._J
D
2.)
C1-
9.
(’1'
3
7‘:
’3
'S
D
H.
»e
O
w
E
D
J
’)
’)
(‘1‘
O
3
n
Q
0
3‘
C.)
a
D
c ..
O
(1'
(D

I.eq11 11" Ed?)
4 4’ \ mu .1. .. . ' ' 1..
(LL..€C]. 1€M£?;. liars ulna: legatith, is an; feed; as 31: tLANSC
1

A ‘1 . ’- . _. " . "‘ “‘ ‘ ; 1:“ .1 1“\ A "I‘ ; ‘ A r Y'V- ‘ '
studies, the criterion can ee reached cefore ,ne caIccta-

tion has b‘mfll ma Xi: zec. Lang cues are also conti “UOLw

\-J

L.
J

:‘f’v
/'

with the frustration event that occurs on incorrect reversal
tricls. Thus the withdrawal reSponse is directly condi-

tioned to then nd can readily yield an 022 following over-

Since the vigor of the frustration effect is directly
related to the size of reward (Ansel, 1962; 1967; Bower, 1962)
then in the small reward groups the aysmptote of the frus-
tration effect is lower than with large reward. Thus, with
small reward the facilatory effects of overlearning due to
increased frustration did not exceed the retarding effect
of strengthening the original reSponse. The Opposite is
true for large reward. In this case, the frustration effect
was maximized by both size of reward and number of trials,

so that it became the potent effect of overtraining.

AOflaomnmedaa was» we Scales.

H
"-4
r4
~43

a \

d

C)

k I

Y I

G)
-4 )

“4

M

:6

i)

0)
r1

C)

h

H
«'1

(I

‘4.)

5

O

r!
ri

 

.34. -
v»... (Pl—I
7 1.4.d . ,. H.a m.w: m neamm woes 3 .
Q\ fill Pu. F4 m N. w on m 0 Q 40.. ..lvxu .. kin-V11} H. \J x. J. ..\ GLGHWOCH n... CHM“ p:
mo. V...» .2: (Seem ; rm larcc C r\... 3.3.0.51
m.mu.qaetp a o.m H m w.m ea c .moatteato -4 a-4u-m.-u
: ¢4353V4 m>O.4.44_..4 NNN C 0 Tucmlirswmm pm.m.w OJ mHficHLP ..rl w 2.4:. «$053....»
(Or.l\x ”KL W1+Crunfsr V II .... I 0

 

.mee
.\ ..l. ..m. 64.1.0 Tim POOL
o o . AV Pl PH \ . )3
m S «M\O.Ml4_ ICHOO NH\OH Tidit C?.MWfl
CU. IL .\ , - - J
dogmass>eo .moflscpman €n+u pecan
.. . 4.. .13 3., I144...
44 - 1 1) . Li: .4 i \: .3 ,. O a. m .3H 4. 314?? scarier
I a c: we mam.rm-< m .r m: w m c o\ \ 04 rc re es 4 .
o 4).: .
LVPINm-tx.
C O 7H 900%
El. .3 lemOU NHIO-.r .menh.®..,..m.t. rHQt
52/ Cl .% W. G :1 2 1 \.3 3 NJ: 0
41/ \Oo. vg “(NC Wiflefl-WFFQ CKCIW'L 4»- (.e ..r
\q | t é \4. V1.4
. 1 .33.- . \\.\o\:. o a :\ \r.o~ 4x41, \. ON A. .Wmfl.\ \ CC? ..LC/HLJL.
r;e ctapr:l; new «up m ms at m _ a PC m i as . I H 4
134.1l4.44 $4.39...
< F_(L4E(.r._.rv. LC... (.4
.2
13.4.»«.:4-)1q. -3 44: ...n. 11.. ; .4.n. and.. a mmuwmdmpw mrhhhHavdomw
(LC 4<C-_rrlL.,4C...rC ...va r... urukt D L NE p .4... f. ...(r.... .

‘I Jul

3.... [U r... CL r
Hospsoo amaseeflsegxm

seem-

 

 

... I. In. . a 1-
made asamemxe HH macaw msflssseq Hsmse>cs mo anaemoa Has we steazd m uh tqc

+3
‘— 4
v‘ ‘
E1)
Q)
H
C)
34
m
0
.3
,J
O
r1
r4
<1

 

.1) \I . 4-
_._ . V ..1
( C
1.24.) )4 )3..- ).)o\.i) .. o\. D). ) .\o\ .0 Z .\ ..J43...)4
L.....(. (.grflpf...0.§ ..CCC. C111». .:. C CC \. C C 4» NC Y Ti. CETLS
dganoh Hawdu Hflmfim a mvhmfl
)4.)- 14. )o 30).. \. 4.. .1 J. 9.. 4.1.43.) 4. 3.)
CL V». C C C U CCUCrtHrF pCQH {(wa QCrflWrrrx.
1.41.1 . l4.- 1‘ ‘ I. .\). u \ I. )1. \ O.) \\l - flIV \ ill 4‘ .1‘\.. 44 I. ...la‘
FLO 0.? ma. ... v0, .. .\.C( o 0H4 C .2» C W. n TnHCNVOFH -.CeHQH On. Q rGHSM b. #56:...” . $071.4...r11
...,...H....J.......3¢ .3614
I. \ ‘ {1'1 .44.) 4414 II
U THOrHOd FLO/U. v War. CHOCrrer. 0»..th t...m4\.0».m ”rem” (4.4m
1.444.) \r. a .1.) ..-Ufid) .4. u 0 \ui. O . V O 0 \. O .\ 1.- .J “4.4 Jill I ...:1 4..)“444 A). 1.14.!114
rig 2:.41.c. mm; «I» s a c w m m u m cf 0 Q4 rmuwa4 bk pfirzflkr::L

1.4.x. MN¢E19 .HfiOHnfiHB

 

11.4. 4‘4. .1 1 .114.
1.14 ...«A .4. O 3.1. .541 104 74.4.4 .4. 1.11 .43th4414 #14 4.44144 wan. .1».
.... ([141. h» C L r1/.rl...» 1...“. (PE L .r 41 rLb4r4L. -.l...rL.IHkrrL

4 14 4.4.
...IrO Fir 4|.L.r.

1.4 .449.) 41.1.1) «I. .1111...- 0 .0
DoncflkuiiHU %O EUPE Q

U)

HOLQZOU asymmfiflgommm

fiDOHU

 

 

4.1.4 1.4.4.».

.mpzmuflpmgxm HHG 809% me31604 mepo>om mo mpfifiuop Had mo hhwassm “9.:oov11.m 1 v.5

-\ --v-1'“.“ v -- a v m 1' .5 5' 3+ 132‘: r‘ ‘1 317* r:
(3..:... 13.41111 SU.-L..'11.’.L '1‘ 11.1...) Lia's...) u, VC‘-.CLU~.LC.. ...2
' 1.‘ ‘ ,‘ J;— vwfl ‘ , . ‘
Iaole 17 presen s a man ary of tne result obtained

for all five experiments. Experiments I, II, and III in-
vestigated the effect of overtraining on the subsequent
reverse of a kinesthetic discrimination ir which most
cues :3re irrelevant. In Experiment I a long stem an
small reward was employed. An Analys's of Co‘m iance in-
di cated that overtraining siC giificantly retarded reversal.
It was hypothesized that the length of the stem may be the
relevant variable.

In Experiment II the leng. th of the stem was shortened
and large reward w-s used. The effect of overtrai11ing was
not significant when the results were submitted to an
Analvsis of Covariance. Thus no ORE was found. In order
to isolate the respective contributions of reward size and
length of stem, EXperiment III was performed.

In Experiment III the stem of the T-maze was the same
as in Experiment I and large reward was used. A signifi—
cant negative CEE was obtained when a 2X2 Analysis of

variance was performed, with one variable as the speed
of original learning and the other variable as the amount
of training. It was noted that the greatest retardation
due to overt raining was for slow learners.

Thus, it was concluded that the length of the stem
was the cor mt olling val iable since a negative 0? E was ob-
tained either with a small or large reward when the stem

was long. It was hvlaothesized that a negative GEE resulted

C")

\ fl
10

nitiati“” stimulus for the or1'1ral reioonse
(5o) that ltd to the ”oal was plac en tefinorallv ail spatiall

1° 1...1_ 0., 1 -.L., ' U..-4— r111° U
distant »ron tDC Site of rein forco1. LU. inis occurred

. 4... ' ‘U,-‘,. , . 1, -.._ - '1 (1 1x
b3caisv ov3r-1aiii a Servel to :lOJWQUc toe s-n ch31~.
T . .1 v V t 1"\ A u\ . ..- O “A 0 a J- 3‘ N ”a . 'I"
Ehus, she; the aninal reversed, the lultlgp;gb SthUlUS

_\ ..~ ' 7. ‘ r . o ‘ ..l—V 'Y. ' o
for lo was not in close temporal coiti uitj nlth the Site

,3 ion-roi11 o1C3“1‘t, i.e., the site of roinfor ement dur-

ing acusuisition. This seraration of EC and one for 10
.L.
Worl‘ Flor ”own the conditioning of he to is cue aid ten:

1; ode the acquisition of reversal. Due to the lack of even

traiflios the 3.? chain for the control group had not each-

"11-

1 '-- — n - J—" ' - ° , i . U . .
CGHIAOL far f1oq the Site Oi reinforceuent. inns durir

1
l'

as readily

conditioned to Pc resulting in faster reversal learnin;.
Tresufiably, shortening the sten 1h1 ically limited the

L)
c 3.-
:3
O
W
9
C '1
i
O
:3"
m
H
C)
C?
5"
9"
U)
0
<:
(D
F5
cf-
"3
Q.)
S
1;.
C C?
:J—l
Q)
:11
H
H.
C1-
(‘7'
}.J
’3
Ho
Pb

"l w. U 7' ' - ' -1. 1- , 1 - - .. YO .-
- 13r1nents IV a1d 11v3st1U3Ued ti ofleCU of over-
. 1 A." . J- ,1-, . ,1 ., 1 - - ,.
ra1-1 u 0: t13 r 333al of . .Tre 11510-33 1vP Lhich all
CUOS were relevant and reduedant. I; Exgerfixent I? a

n.1,...11U U 1- U. . , - . _
shunts 3d Sued was used and all S's were given small reward.

L - -‘ - ‘- (N V - 1 . ~33 . .-. ~.u . 1~ . . r n V' -~ I‘ ~-‘
It was beLd tflwt overtrainiu s1 nificstntl retarded r3-

0 L.) “‘d
«. r— ”'.. Y'j1.r I ' ,—.- L" 1.. ., .\, .1 .
veisal. in E-n:rihent V botn Sine oi reward agd amount of
J "I“ . '3. \ ' N , 1" ". r“ ’- 3 . f n‘ J‘ r A ‘ L
LZ<1l-11 (J 1 7‘. n_l£jU];1In~d 111 El (1?? fEMZLMDPl 1] (q-f‘__,-. :1.

was foU d trat for the s all rewarC group ovortrainin

l—l ”,

J)

u» \u'u .-

{-‘v‘offf‘!
'-.L ..

I. (.4! If

p ’U :2‘

It was

”Vtru
‘

V‘L -.

"‘51 1"”!-
”d

1'

“NC
.
' v‘-g

1'40
.5..- LK—Lg‘.

I

-.,

(.3.

”x
.1

o

‘1 (1 ‘rx
"5 V\

0U.

I

1..
i

1 C

T “\

elicited

a

27")
Us;

‘11

:“Yr
{2*

. ,_ ' - l- J-
,V \v AVrl“ 7" V- ‘ S
y‘al‘ql. —.A:CJ.JI.LIQL1

t3

,"\
v

1:;

3": S C

J

y. p. c w-j (
‘ y VJ
‘-

.1.
the

of

testatixre C...

...’.0

«Lawn
Ul' "

1.

1°88}.

0 I
7‘ “1 7'\ I'.

Yl'fi 1. ‘ ..

.g-

A

flu 115 over

tory

El

0'3' [-1 L“.
‘v¢. 4. ~84--.

l 11

I.
.Aqu

v-n 1'}

two

nr” (3
n. 1,)

1'30

if? 1" e

wt
'5.)

v’“

_ £11.,

1d why they are

.I,‘
..J.

3.

is

16 has

1,
1.

t

O
"- .1. Ci“;

3!
«L

1- ,_
U-l_. VCU'

>rie*

\—

f postulatJS thr

ion Theor;

Elicitat'

L)

1 r” '
In: a

1 if} T'I‘r" .

0‘?

I
‘0 1"1 r
Aafifigtil

tior

i

Citid.

[T114 (3
. -h {s - ‘~./'

1

‘r m .
v _ (_)
‘1 I.) \I

.1) : C)

"..r

- ."'\ r‘v r

‘

1,,

(DE/3'

\W’)
‘-

f‘u ‘-.‘
Cu -C£J

"ent)

0‘.

1,4,1 m
a. s.» 1‘,

rfo

O‘T? r101.
“5 as ..I-b‘-

{‘14
‘3'..-

T] .7.

(73,6;th
‘v‘ vi.

to

U, 51

('V

irdgzg-.

o
‘I

’3

I)

4 1 ' m 1 --. 1.
V. ,p— a'- 1 P- D 1 W ‘:l -‘ .‘y' w. ’1 1 '-‘ -| ‘ ¢ v , ‘ ‘r); V," P}._T (3 L]
..-K.‘ v k..'.....:.‘ L“-‘. .I. (. ~J...‘. .gfi—u. .54..» sJ -. u‘ ..L.‘-‘. a. ....- ..U a .4

§~d

L: i. ' I"- .. -~ . ow 1.. .1, ' .1. n: 'v _: ,
Sblmu1l lo: 10 far lies the SiUe of ixustratioe

' . , - U: .._'.. fi .. .' -‘,
curing reversal, retardilg the Conditiohiug of the with-

‘w-J— \ 1' - m t" - -—-....4.-
Green lUs, uou a ilcrease ne CllCl iho

" .1 ‘~ A. f‘ ' ‘;‘ "' ‘V‘o' .' 1/7 r-‘wtfifll ‘ “ firs-v arC‘r) '71:“; t“ in a 31 fl] lfilr t1". ‘.
.I. :14- Juv le'f ...:-OA~-& 3Y1 l'3'.~"-‘~ ‘-“- Ly-.. l"'-L_) l \r i ;..\.L' 4‘1]- . *AL‘DJ 1ll“_‘1\yL-J_‘_L (v4- . bl; '1-C}

J :1 r fir r‘v ,-_ (V . P'J ' 1 4'1 1 . . '1‘ ' j ‘rx A
1_U:1:e 31 It <x--e b0 or e: U CCLLJJJUWWULJ CllCleu.Lwr-AJu-

U. , .1 - . p : ..,,. 1 - 1‘ 4- - -U. z ' 1 r .U .U -1 :1 4. .
resail resulUlhD lu UeUter COLdithJlub all --ster re vc e1s11.
-‘-- - -. - u T ~ , "L" . J- 1' '.w n L 1. '. . .
I; Exyerlaents I, Ii, and III bflC first euplathior is

h . r \ V,‘/‘., , - 1 ' ‘—- '. 1 v ,_ .j
applicable whereas tne secolU 18 hot. In these erp FiHOLto,

{u
m

:1"

‘<

the oely we; a subject could possibly learn the reSponse

r‘ A!
190-

He
he

through backchaining. Since the stimuli IJCPG rurely k
thetic, "hey could occur only at the choice point and before it
(those kinesthetic. stimuli resulting from making a cen ri-

fUQJl swing). This leads to a condition xzhe re the discrimi-

I»
}

9.
{—f-
i.)
4
(J
U
(1'
S
a:
k...‘
'40
m
f
O
P
.J
0
Cf
O
O

htiguous with the site of frus-

O

trati01i duri;1‘ reversal. Overtraining serves to heighten

this effect by strengthening and elongating UhC chain. The

number of learning trials and since the aysmptote of the
frustratio on effect is reached in about 35-MO trials (Ansel,
1962) overtraining would not have this effect. At least,
this is true for Experiz1en :3 I and II. However, in Experi-

ment III at least 6 subjects (3 in each group) pre sumc bly

reached criterion sufficiently ast to obtain SO1e benefit

f2C1:1 O’”‘”t1f1; ii;,, isucrrrmiih;; th" frulstzw‘tirr
a :"9 igovar was prrforuel the ae“ative CEE fo
loar"ors was rot as frooou-eefl as for to: slov

In Exririwo ts I? 1 J hath the above e"fl
{N””llC?Eflf? TJIT“QflCliJ“ the Trrnher' :f trialjztx

.I. A. '~ "' L)
the arnroach “:erLSe axi recoiel" :ause thn

1- 1 -
hi estietic chai‘. this vow“ rwsult i; the n
In Exfe iuogt IV usir; shall r«ward the Q's at
Urigigal fliserifiilation in affroxiuately 25 tr
$1311 rerard Prov~ of T:”(r113 t V original ac
occurred 1‘ "inrofiifiatel‘ 32 trials. Both the
‘1re (flosex'tt til arsv‘rtotr <3f tfiie irwuiUrUJLioe
there is little beiefit frox ov3rleainirg. Th
green of Exrzri‘eLt V acquirgd in affrO‘ gatel
This “otud 7 av: considerable rCOe _or berefic
in the elicitin: value of nor-reward. I; aiii
reva““ J‘elfls a still ‘reaxer iwerease in Ten
elic ii ”rust: tion-mithflrawal then s all reW
both c0hsiierations are combinei this "ouli o
retarflingo e_er of overlearnins ard result
0-3 .

Tron the above analysis ore Cal Sfecify a
iaCles speretive in t‘ learrin“ situatioi To
"secific result. This is WNJike tqe stand of
rftical positiors in which one or th faUtors
to yield C23. in exafiblw of such a t‘ooris'ng

leans to he

[\ 1' (:1 ‘4 -
.- .. . . U
. fl 1.
.1 1 ‘5 'v
1" — ..

' o

' ’5 1 ‘u

-1. \‘,'l-'- _L_.

t

O
—‘
u '1-

’". rij.

‘ J—- *1: .,.

...::le ' ' ’
U. i-

,. U,
ounev
’3 ”(A C‘

.. 2 k!

‘. (‘ J""1
.Lo b1:
wredi

-reward

1.

- ,5-’
{11
1h

r3'

0

ials

1 ‘.

Lif'il
1- -1
ell"

itiV'Té

‘
l
L

V,\ v— ._—.‘
1

k. .. r» ...J

 

1 .1-

3... , ”v.1"

- l
l s aha-'xxduat-

o I 1
1 “‘. "'- .": f") If?
.A—J... Ll-..v —'
a
r?» ' If”. ' fig 1"?" . f ;
q u. w -1-.. . i"- ,‘_'..a.
9 U
-_.‘" “‘ "v ‘7‘0—1 - ’_1 . |‘ a
' _ .0. u....~..(..._.k/:J

m
y-»
J

/,3

$3

)

, .
I“.

*4.

un-*

O

:‘:)‘I
\A-c-S—A.

'T'

7“, v—n-v’}flr‘ "—11 “ fl’fi"
x - . A .

l
A‘dg. ~u.~J--NJ_1~

 

 

 

 

 

 

 

" 1 '1 -~ " ‘ _ a“. ‘9'"; ..,~ '\ 'i .' '1" ' 4"" "1 (‘1! 1 1 "‘ >1)
-.5(--‘ - _. ._ .- -.l. ’ -- . - , . ~ 'L U: .- -.1 —.-. ’1 f .4 _ -a -L. L/ \J l— (3.1.; LLI bk .1 b_-'v/L'.1_CJ_L 0.1.
' ..v- ‘ f‘ . . ..-,-- 3‘ -. _ .' ° ,3 1 , - p... . .3... L. .' .. 3‘; ° ‘ f .1. ' _
'x ,r‘ , ..~ I \ . f“ . ,\.f . a , ,I‘ , . f 1‘ - (“‘-
l' ._)- ‘--‘_'- ..."U': ~.‘ (2.5-x; 1" -‘ ...:- .1 NJ:— +~ .‘r v uL ‘. v.:./ V. y; \t' J- C‘L ’4' \-'J—.--b‘~]. w". L’l ~/.*
U 4. I)
r;-3,q ‘,r , 1 1? .- f3 :1 ,.\ K . ‘1': ., , 1" .2 ,‘1. , g] r,. 33 :.4 .-.. {a
'. -- -‘~.L 1“: I .. .... C {‘L i..- .. O}. V 1‘ - - v Q g... ,- LuLJl J. .‘u ¢ -t? J ...‘ D; L; ._ Kiri. O .—
.1 ‘- . —~. ’-‘ - .1. J o _ ,.- u ‘- o -. . C‘ L 1_ vv 0 ~ ‘ .‘ ‘ 4- 1 F’ .
A ‘ fl, " f“ r.‘ ’\. ’I l‘ _‘ f j n -'-1 'r . (— . ’3 \ ' . (..1 ,
\rLCC UC- a].- -‘ i'—’¥J -. J. U—vU—Ly)-..’ ‘..Lvl-‘-D~—4a‘ H Va UK. L-:~l I J- 'vl LJd ’ // l.
C
. -. o O I a — n
.\ 1n»1 " T .~~~-4L3.n-‘ 1’ 71A .rvr‘f'. .) I 1 -- 44 1 r 1 ’iffl ' ’1") ‘(z-" __7_‘.’\+‘
4‘. 4‘4- ‘-.\—L. ’ A"; . J. Lu- —/ J- ’0 LI'J- J 3 4-- A t 12-: f 4...}. 1.;Ja ‘v .Lkawu l .o Joli [‘1‘ Cl ...C/L-L J
any] 3:6n7p4 ‘11" ’1“? '1‘ '1 , ’ vr. ] 3 . "‘vaw 1-000": ' 1r4c‘f‘c‘nv anl
La...L‘-- ' .L-JV— -——..-_.. .-~ L'-..L-- ..-V -4 --.A..._L_) . l4 '. _ .. :‘J \4--- -—Lv.) » 4; J L
a 0 ~— h / A / A
'1 +,1,~T.q-n(z+1c"." ~"f‘1‘f11, (‘7 " find-1 1" 10.49 ...0 1A ._'7,n
L.I V‘.-V\-- v v... va. v.1- JV-..“ ...‘g;-. _ '~ U - - ~4-. . _- . , -. I A‘ , - :’ J J" v
5‘ r‘,’\.' A ";\’34 ‘1 '71 11”.}“? Wflm'fi t nff ntr‘ O 1 f. ‘0‘; - I Yr\}"
A.Q-.A ‘, —— , -4. I --v --.‘-~L-— .- Jot-AIL— O... V \_ Cl.- V CV is) -~ \1 lo .3". W-L.— _. -
("11"f‘1 '2- 7" ”#1 'Ith‘ ° 1 P '1 )C‘tY"l+iOT’V t1" ~"‘-f db‘flivrnl f‘r" '
~ .n—k) U -.V..\J‘4 . iL"-\I L - 'v _, L) 1.3.: J» 1 L55) .5 r/IV .45. -.C v' ‘ .- "a. y x - O“.
». . ' J- 7 _° 4» 1. . . A t —— ' ~° 'P , " ' ' r‘ .
(.1 WI '- m, ‘ V .‘ n I? .v... Y Q . .: » .1.-
by J I.) .1..(\ 'gld OJ- 0. 3.. Ugli—A-.) bl LIJ [JV L; (nu—l C ILAAlVA- UN! . i-l 1‘ . I. . IJ‘ (..nce
Q ‘ ‘(3 Tn "‘\' ;‘\+ r” (‘1 ““ ".11" C D fir.) fl rn‘ll- A“ “I“ rfijw. ’3 .1 A r-- ‘7 fl p 1‘ A»; (1 1-1’ \ w. ‘3‘ r3.
Va... .0..- - J J... g z...) 1].,» , ...J. 2., .4“ >3 Q ’ .. --. ‘ A M x. .. .. 'L)! , .. _.3 .. .-... . ..A.L—J
v o o ~q- . f"
' (:1 1 ‘ "‘ r I Vq f Y 1" . I l . (W ('I (I. ‘
1.: 0-1w1t13n. LLB or». AC” 3m 10 re , 1,3(.
‘7 fl? v in r.‘ 'T n‘lfi 1 \ f - 1" f" A». ,-\ ,- 1 t . |’\ (‘1 . a,- _ r r 1 (—
--.IP-, 3.3” 3 _L 11m,<:;‘3f reflcfl‘rw&lmc.lohn 13-:reuaru -wa
Y fi'v" Fn‘fiffw‘ i‘v':*vo't3 IWM 1 tH“ "‘*‘*+Wflc-ri” t?“
ul- .- ...— -— A v v --U _ 1.- U V-J'; — b - I UH“. Va. 1 -- \. — . 3.4 /
x~3q . ~.1 - 1 h. - s— T . _~ -1 , ° 1 1 1 4 r3
I 1 f1 ’ I ' ‘ f\ ‘ v ‘. ‘ h 710 7“ ,fif. I z
2..- 9.x; .vL 'LWC'A; Cl¢'.)Cu. L o 0‘. o 1‘.” uJC‘L. .L‘ d v- ' ., J-’ v3“,
-
rt :03 aéw
4‘ ) ./ ‘ ’-/"’l .
nmn““%fiwm :"TT 7nr7~ T w ' “Wm? r"' -“~»uw771 q~a
.773 . - .. -J, 4.-.--. ’ -- 'V' .1: . w H - ....’ V cu . ‘- , x.-- -
L77 7377 777~ a-« .nwwnwawu n"\w was sw7 «w wa nna
_ _ \-_ .. k _ \ x v- .- hfiLU ‘- -_ \JIa-n-o O ‘— v -1; -L‘--~-"-a—U -‘..~. L y" .’ J_-'_,
(‘11-’ij1‘ . "'0‘ w‘onh'] a,“ cps-17":Yaof'1wvm "v 4 1'1\.*«' '; n1 (‘\"“.1 NC".
"a -.-: . V y- . --l.. v.— I ~.. ..- ,,_ L'..L u ‘ ~.. ¢- - -- l L... __ _' '_L_)-A—v .-
T‘, I’ernj‘l A1 («I ray._ 1 r. f)? {’11. r} (" (\
_. I—JJ K. - _7_ ‘1‘] ' . -- I v - , / n , / \. -_. 74“! .
- ""- n7 ‘— "1"‘/: '1 AV: ' T {‘1 ' C‘ (“)1 "7 A 7 '7)? .1 HF“? - I 71"“ '\
~- -- ‘- 1, . —-— ‘.—L\.~ ’ L: .-.o’ .y‘ k—kLa‘k_-., ..J.’ V- nu~4r~ .4-— \’¢- ‘..‘;. +.-_

overl3arn;ng

 

 

 

7
., 1 ° ,. r
and QTlVG 13 J01 in PCVUPSJl
3 . T m 3., -. hm; rxw : m:
133:; go. .Ourgal of C33raratlve auw -n;s;olob;cal
Th‘rr-I Wrm-Wr 4’ "W F14 rf‘r" If ’2
_:, -10-- gym 1. >/\., 1.1., UU{-\J1J.
('1: *1 v \, "'37.”.A‘ . r, pp . '- ., .13,
133 3.u. Ovrrl &;L1L; TCVUTSQl e LCCt 13 a 5p3333l
"“ N -'a. l.-‘ ’1’ .‘Ag L’ 1, I \Y‘ . '- ’ T.‘ (:1’!
4LbClelLaulO. deL. I3rccitm31 and Lotcr 7“;133,
1! dq: 033
*U, dl/Jg-J V.

C C‘ 7 1 T‘)
131, ;.J. 3 gtcvonson, L.H. hes
' 3- , ., ”m;- n .- : .° --
dlflereut amOuuuS c; trainlno.
1‘“ “/11 ': '-‘. " T‘ ‘IN -| w~‘
a»& ;L¢3331031331 :sgcu3;cgr.

 

7‘7 -'

“.1- .
J U041, .LL...¢.

learning
.9 Fur.-. _
C'Il. LJLiJ‘.‘ 1"

RSV crsal p
with and

W
i flfifflj Ffiv
..A-V‘._.4\-i—‘_ ‘ j.)

rforzu aznco

CL}

3 reversal following
7* 1

.T-r) ‘r,(‘ - i-‘JE?
odva- Jr

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l O O
”ml an L, ,“- 1". '7. T_ mum-‘50 ‘L. 5";*.r:‘."1 ftp-u 1554‘ rt‘:‘.’1._T-."5"i "1' f7?“ I‘C‘V" ”-
d—v ..L; ‘l-U I - , .- .-;. w L‘IJD. 'v-R—‘l, L.. ‘V l . —'—....- -.- K \_: Int-p. L—b-—~-.- .I-L
. / /
('11‘1 Tm11~nnr11 .11," .T“—1—~.r-\.‘.')"5 '4.\‘H F01 pr‘flff‘ ”‘01 {‘5""-' 1 (3530 (1":
~ _-* . . I, - . .2 _/ _'_ L , {...
5.5.“-.. " ‘ -“ A “ "‘“ ' '" 1 L214 I a 2 ,
4 "V7 '1 '2»
LI. ‘- L’_-.‘_I.
.- ' 'x- A- ,-'. m - .1 _. ,5 5— :1:
3'4”450 ” 7 SO“??? “. LiwfijOQPrlwl 4 b?l nt*nss n+0-
-..: r - . , A- .-— . u. - w» o.~ ’ a—u V H DJ 1 3“ 1 1
. _' _L_ 1 v - , - ‘1 5 ., ”-2, ...5 A. r, —> 1 -r\ run, a A («.17
'.~v \ V) r 1.1“- . ‘ L1 54.4 . , . . ,( r "5 ‘
CI -... .....L’ -./l(3.1. L ? ‘.‘ {_-_' 5.12.1. . r; n‘ _ 1 ()f ___‘__:.A- I :l .51 L 5 ’.-“_ 5 _ 5". ._ 3 - ’
/ , ("5
146d no dvr_4-5
Z / 3 " ’ .a‘ l ./ .4 .

‘v—‘J ‘ 5‘ ‘ 1' 1r— Tfi T 4| "“1"! :fi f? 1 ~' 5‘ :1 7.5‘.A‘ '1')" ' § 1'"? a,‘ / fi-n' \ (‘11 1‘31‘) {\— —" t T c: $.11th

A. "“0, ‘_o. . .Lw ...“ -'<' -u-- bo--. . u" m .- - \...~~.j, wv. . _.

.9 ' . 'T' -. ~- ’w 7.2 : - - .- v D " ‘." r ('1'5 ~~ ." ‘ " (EMA
J. .r\ .A ‘4 j- , :fiJj— _. :xfi 14.1'5-2‘LQ v ,5 _‘ -]-l \ aSl t r (3.1. 1,9L‘17C1-flj55'1. . '5'." , 1 ' ‘1 . . .

fl. ’5‘ - " 'fi "\ 1 '1‘": 1' .. 1’.’]'1"-" T -r n Cry‘f’n”i (‘ '3‘, :5 2"" f1 _f"+“uf‘.‘fb

J", u 3 -..-.. .23. --_ ‘ “W‘ (_) " -*‘~ 040 ' -~ -“~ U04. 5.5...) ..5-
(m7, {‘3 \ f?“ .‘N (”‘1er : H+V117 P50 ‘1 ”\Q'n ‘- :5 “‘5 -" v".' Y 1,0751’ ' 7‘1 “‘5 ' 1"‘f T‘7’351‘5f3‘f“
A\ ~‘ i . '1 ’ _ 5 A’ _ I 5} V. _ I \, __ _ lg, ‘. V, .1 _. _, - , - . _‘_ .. .... ,A ‘ , V.

."\ /
1 ..‘5'7
_ I x.‘ .
. ~' I
“if 5 ~55 " “ - 1 f5‘-r‘.+ 15,-}. tanO-nif chat“ (r3"1‘*\{—_5“/11 "(Kg-5 -:~‘ T T “H T" .‘szjs-ai l
1”. _U , ‘_ .‘_ . _ ;__._. - -_' ..-. -.L‘J .. d M.- .,'.-J . .A ._ 5. ,Am .r , ..g... A ...- . .—
'.“. 1 ‘5. WW": 17A q—mfl 1 rafivomfi. "5 "‘ .?f‘.'-5T 1"3‘71” "’15‘3V1 f" ““5" 1‘"
—IJ;.—- . I , _ I‘_ 'J. r -J- I _ C":- .' ~—:—A—_U . .L. JUJ ... -/J i. . C‘vk. ~_'.-L—L \l .'J Kr ’
1 (fr/50
1 ’/ .
‘5-1"F"’\1”\’7 " fl ‘X'lfifi-l 1n ‘fi"5.‘."lf~‘ :5“ “7‘51“!“ fl '1 .1“r.."':507 0.0
4". 1-: CIA/LL' , IL. J.‘;*;V¢:,l 5:"... _ 5,51 5.1.15 “.15. - :4 J!) r. 0 c V 5.) A“ H. .5
. -.‘ p .. ‘ o .5 — .-5 -- t r —-- d «I: 5" 1 r‘ 1‘ "‘1‘ T - ,
‘ ’5’“ - 5" . '1 {'3 .-’_\TW ,- b5 .1 ’3“.- f‘ .-.) - 5 ,
T'BZ‘L;"‘.L-'q. G O; fplmfifcg. 09.3.- 5 L 5.1,-u. ”‘09 5. J -‘L 5, .-- “1J‘1J.
.5.
*fl“'»’1 '41 (51“ :"wpnivm HW-O'I T“?! 75"}?(‘1 {\I-‘jr 1 O «'3 54K 911'..qu
‘ ‘ ‘.__. . .‘5 - a L -.. x- 1. \l' ‘-._.. .‘v‘y \J— ‘.. _ I E ’ .h / ...v'l/ , .1 ~J ’ ..x . / v .

m ..5 L, 5 j. (.5 ‘r\ (5 ”,1”. (‘1',53 ,\ '5 ‘ :"J‘C' TN 4"
(11'711nln .1. I L: .4 193:1" .. 01.- o ‘v'0 n "- NIL-L13 " _— 3L ' ..i.. 59 l UK: O I-"- ‘ o- . O
~-.~ '5 55 T" T'T‘er- r“ “fl 1 '1‘ V ' we" ' M "I ‘-‘- “'2‘: :2
...?” 1 2‘ -4 -.. ....) 5.19.3 . , , .2 5,5-..4.:5_.r;,o 1;; L-:‘1t-£19T5.wl-a..‘.c'.
..,-” L5 w‘ 4+ m-.~a. v ~vpnfl'5 hnwfi” 1mr¢

3,111.1] 1,140 ‘Lro LJ\J'-"f.{11‘l v-_-L.‘ 5...-.5.U.J 4.;L-5—J5"’ 4—///,
N. A: FT nr‘r‘\
'!|' 1’ - i- ’*
L ‘. . v u/ ‘H _ ’. .
Tc: .1 '1 v ‘ v—j (:Qfifi '\ ‘4 If -r"‘1_‘\| 17“ ‘fi‘,"‘ ' ‘ o . o 7' ”)V31 '1’“. f- ‘o"_ ‘ u ‘1‘ ". ‘1
-.y- -..L, ..r..._" ., 54%;" 5, ’ M.I_"_.H.. ‘;a-_..;_.IllCé3.blO.'l 0'. OJ'VJ. J _(.‘- _ 1..---.
‘
' T . *‘5 . .r. J. —‘| rm.” 1- 1 .,.VV
-v ".‘ (fin-l ~. V1 A11NQ‘-n1 51‘rx z- .5 5 Ir‘. A 5'
g. ' ... 111 a T-..a-o. 5. 455* _. 4f 5;“ -. -- . _. -u 5. 5. A .,
40/" Kr? 74"? (‘-~\
..,-_,.., -/a .1“"\"/.

-. c < -v 5’ ‘ ~ ‘, 7.. ‘ . K I > . _

5- 1‘ ‘ '--' *“ ”74mm" 1“ 51; (31 LL 'tr“ 5 J ‘T . T-L.‘f5:: zwirwz‘mfl Often

..-—w- , ...- a, U_Lv~' , -cc-:Jo -- d.“" -O- -F n

_ . O
P‘ " '11 ' D ‘5') ‘.“ r, w; ‘W="‘“- ‘ ‘I’W‘o . I,‘o‘5‘(1 3’73 "\""a‘55.. h
4* w'; 0.53: 3‘ 4.-. 5V 0‘ x , w Lu 3,. CC ~_‘ - . ct. -- A - 5.;
.-. ,~ .’ / I .

‘7r ..-. ,« '-‘! ‘. "n1-.’\1 ».-.. «LA ,5!» r°t’_.<."'\
“-.. ...-—. - ll 4‘ _-... ‘J._J . -../v..., y ’ ’1’ / . r.

V ' . ‘fi— . 1- ‘T-l‘ ~ p.‘ 1‘ I 1 *- ’L‘c J". " f9 1‘ p rm ‘-1 1‘ I - -~ 5 '. '- ,r-u n 1 I“! “ ‘1‘ v

,- . I. . I 3 1. 5 5. - ....\ l 3. ‘. .- - .
-..x)...‘ u .L , L, . .....- 4.1- i 4..- '..'\.u 1.2.1. K, C .L QC 5.1 CM. (.115 VI .L‘Jgii -50]... ‘U (v.1. ._ Lag. ......-
/

-°-. 4.0+ -= ~~ 4. ~'. 5-5-, 5" M. ww—n tmmnm‘z Nu» 1r: .5 l1.
—.‘.~——‘4 1.x.{v}..- ....\ v... _. .L... 'Q Lid ...)..a. U‘v- _ Lb‘~'l-’ . A ~j1 .2-4.~., -I. '- (311 , .~. I ,-, - ,
r50 “5’
" “-24 0

"r 5, ”r ~.~--~.c~1 ViruL'v'd-jv- L4. : 'f“n*'-f\'r"‘;4 fi‘N’TWr‘JMPfi .. mo.

-.L‘... ...:.v..‘. L . .LLV b v.— *- Hi... as v V'~I.~ ‘.«wd ...— .. a: .. \ U-UAL .L-.. U... \’. 54-x) --..U ... ‘v‘. V \- v-

7‘ "I An?" , “(1'5 .. “ "" f' V:- 351 5'“ '3 "1 .vrm-L'.’ ‘nvj 1‘53 ."‘ ~ '1‘: 1";5 7:3“. " ’7
.--Ol_-r\1‘5-1L-—..L—D .L J._~.L.U +1.1. ‘ Lu AL-I—J. O 5.A-.-— L‘ -- ’Uk’ 0
-"- 5'. r555': 1 r5? ‘= W" 1 '1 r‘ 5"“. .r") ”N 1r) 1, ..‘Y "
- .A~.AV--l_— ». —- --‘——..--1,.. _ - d --\A . u , ~, _
T"‘~—~ , - -—— "" ‘ 5 '7' fl ¥-"~- .,-.L.". 35 '- - 1" ~ ‘ WA ~ -.1-v.~, -.
: -’.‘I'.'\T (1 r t- i 5r . r. , *1 . , 5 1 ..5. ._, 5 ‘ ,

L... .d \J;L\r . '--.4--d ’ ..‘I- Q , 4.1. I.) JL’.\.‘-J Oi U1.- _.‘ ‘10- ll.-iLLl—~ u'd 0.; 3'." C) \...' ... 5.. -2.. 1‘ .L-~U
.'5 7":‘5’15 ‘(q‘ .fi “F“ '—/-‘ I"5-1 r--\ :1 (3,1 1. V‘ _ '1" 1,..E._,r. -.1- (1 4' (‘5 ’10 15‘ f" fi (if? 41 r\ ’\
J.""'V‘l"l". " 31"""‘ (.2 V""‘ *‘V... “' L ", 'l/ J J, ' l .11.

...." ' T v. A_ T.-‘ . '1 1’5 _\. r" I“ 1‘ Y)’\",\‘r‘ «(51 n 1 ‘y . . , ., 1:. I" >.V~"..’-.’|

_2_.- KL. 1.) o... o v. 4.41; CIA, ...1' . .1.-.u'l=—t_:' I - ‘-« J -4 .i-J ;.'---L .1. C--J-CI'.'-'_:1U ‘.. .Lf-L DJ. '3..."

I I _ - .- ~ ‘ ' fl —. - _ O *—
- .1 If; (1.5,1 b if ‘3 r j _‘ r: 5, I r‘ ‘V'I‘ :— 1 p fir. I\Y’ 9. .‘1
Ll-LLJ‘I C~LLkL OJ). 1" -ClC/e.li'u,,i.l.b . ' I -- 5 ‘-‘- 0.. 4‘“; ._ 1_.“\"‘_-_, _=,_',
- /
Tun-m1, 7+," 5"“‘5’ 4, (w, A - '75 ”3’40 Gr)";
4. Id ' VJ...L;._-‘- , w '. vhlv ‘1 \J' \. -‘—I \.~ .

j} L)JJ ' / , ’

T-‘v/‘w‘n- '“ “5'1 ""4 ‘F' '5“’5'\ mi: 5 V‘l’VfiY- -.-' '5 - w". vow" 1 3f? t
‘ _ 5, d I ,_ u . ' H . 4
44“" J ’ u "'d , -.. . ¢i-~21“-1 U ‘ l J 0- v14...- w- I .4. (v C»... AL-L.._O .1. LE VJ. 13a - £1 .— €C .

‘ /
“f5'vl-‘fi1 110-1 “7'1‘1‘ 4. 'J’V‘ 77" §2r7_1 r"?
_ he'd -. -— .' ~./ \ O , I v ’ I : , v I n J .

 

1

66

Mackintosh, M.J. The effects of overtraining on a reversal
and a non-reversal shift. Journal of Comparative and
Physiological Psychology. 1962, 55, 555-559.

Mackintosh, N.J. The effect of irrelevant cues on reversal
and learning in the rat. British Journal of Psychology

1963, 54, 127-134.

Mackintosh, N.J. Extinction of a discrimination habit as a
function of overtraining. Journal of Comparative and
Physiological ngchology. 1963, 56, 842-847. (of

 

Mackintosh, N.J. The effect of attemtion of the topic of
generalization gradients. British Journal of Psychologg

1965. 56. 87-93. (a).

Mackintosh, N.J. Overlearning, transfer to proprioceptive
control, and position reversal. Quarterly Journal of
Egperimental Psychology, 1965, 17, 7-16fWTc)

 

 

Mackintosh, N.J. Selective attention in animal discrimina-
tion learning. Ppychol. Bull., 1965, 64, 124-150. (a)

 

Mackintosh, N.J. Overlearning, transfer to prOprioceptive
control and position reversal. Quarterly J. Exp: Psych.,

1965: 11: 26'360 (b)

Mackintosh, N.J. Further analysis of the overtraining rever-
sal effect. J. Comp. Physiol. Psychol. Monograpp, 1969,61

 

Marzocco, F.M. Frustraion effect as a function of drive
level, habit strength and distribution trials during
extinction. Unpublished doctoral dissertation, State
University of Iowa, 1951.

North, A.J. & Clayton, K.N. Irrelevant stimuli and degree of
learning and reversal. Psychological Reports, 1959, 5,
405-408.

 

Paul, C. & Kesner, R. Effects of overlearning trials upon
habit reversal under conditions of aversive stimulation.
Psychological Record, 1963, 13, 361-363.

Paul, C. Effects of overlearning upon single habit reversal
in rats. Psychol. Bull, 1965, 63, 65-72.

 

Pubols, B.H. Jr. The facilitation of visual and spatial
discrimination reversal by overlearning. Journal of
Comparative and Physiological Psychology, 1956, 49, 243-248

 

 

Reid, L.S. The development of non-continuity behavior
through continuity learning. Journal of Experimental
Psychology, 1953 46, 107-112.

 

 

 

67

Sperling, S.E. Reversal learning and resistance to extinct-
ion: a review of the rat literature. ngchol. Bull.,

1965, 62, 281-297. (a)

Sperling, S.E. Reversal learning and resistance to extinct-
ion: a supplementary report. Psyphol. Bull., 1965, 64,
310-312. (b)

 

 

Theios, J. & Blosser, D. The overlearning reversal effect ant
magnitude of reward. Journal of Comparative and
Physiological Psychology, 1965, 5, 252-257.

 

APPENDICES

68

69

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR EXPERIMENTAL GROUP OF EXPERIMENT I

 

 

Acquisition* Stage

 

Days
8'5 1 2 3 4 5 6 7 8 Total
1 7 4 8 6 9 8 10 52
2 5 9 9 12 35
3 6 7 7 6 7 11 44
4 8 7 6 7 10 38
5 7 9 5 11 32
Overtraining Stage
1 14/15 15/15 13/15 13/15 11/12 11/12 12/12 89/96
2 12/12 12/12 12/15 8/15 15/15 14/15 11/12 84/96
3 14/15 15/15 15/15 13/15 10/12 11/12 12/12 90/96
4 10/12 10/15 15/15 15/15 15/15 12/12 12/12 89/96
5 10/12 12/12 12/12 11/12 9/12 11/12 12/12 11/1288/96
Reversal Stage**
1 2 4 3 7 7 13 37
2 3 6 8 9 14 40
3 1 3 1 5 1o 14 34
4 2 3 8 10 14 37
5 4 6 5 1o 11 15 51

 

* 12 trials per day

.‘I
l\

-% 15 trials per day

TOTAL CORRECT FOR

70

EACH DAY OF

TESTING

FOR CONTROL GROUP OF EXPERIMENT I

 

 

Acquisition Stage*

 

Days

sls 1 2 3 4 5 6 8 Total
1 8 7 7 9 9 10 50
2 6 8 8 7 9 9 12 68
3 6 8 8 8 9 11 5o
4 6 8 11 5 6 11 47
5 5 7 8 5 9 11 45
Reversal Stage**

1 4 11 14 29
2 8 11 15 34
3 6 7 1o 11 15 49
4 5 12 15 32
5 7 6 9 12 34

 

.3’.
0

3L .".
n l\

12 trials per day
15 trials per day

71

TOTAL CORRECT FOR EACH DAY OF TESTING

FOR EXPERIMENTAL GROUP OF

EXPER IMENT I I

 

 

 

8'8 1 2 3 Days 5 6 7 Total
Acquisition Stage*

1 9 11 20

2 6 6 9 12 33

3 9 4 7 8 12 4o

4 6 8 8 8 11 49

5 5 12 9 8 11 45

6 8 6 7 12 11/15 12 56
Overtraining Stage*

1 13/15 15/15 15/15 3/3 46/48
2 15/15 15/15 15/15 15/15 15/15 12/12 12/12 99/99
3 14/15 15/15 15/15 14/15 12/12 12/12 11/12 93/96
4 15/15 15/15 15/15 14/15 11/12 11/12 12/12 93/96
5 15/15 15/15 14/15 15/15 12/12 12/12 12/12 95/96
6 15/15 15/15 14/15 15/15 12/12 12/12 12/12 95/96
Reversal**

1 3 6 9 13 31

2 4 8 14 22

3 3 4 7 12 26

4 1 4 9 15 29

5 O 4 10 14 28

6 0 3 7 8 14 32

 

* 12 trials per day

** 15 trials per day

72

TOTAL CORRECT FOR EACH DAY OF TESTING

FOR CONTROL GROUP OF EXPERIMENT II

 

 

 

Days

8.5 1 2 3 4 5 6 7 Total
Acquisition Stage 4

1 7 7 7 9 10 40
2 7 5 12 24
3 6 5 4 11 26
4 7 7 11 25
5 6 8 11 25
6 5 5 6 8 7 9 12 52
Reversal Stage**

1 11 9 13 33
2 4 7 7 13 38
3 5 13 18
4 4 13 17
5 5 10 14 29
6 3 8 11 13 35

 

* 12 trials per day

** 15 trials per day

73

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR EXPERIMENTAL GROUP OF EXPERIMENT III

 

 

 

Days
8'5 1 2 3 4 5 6 7 8 9 Total
Acquisition Stage*
1 10 10 20
2 5 5 8 11 29
3 8 12 20
4 7 5 7 7 9 9 11 55
5 6 10 16
6 9 12 21
Overtraining Stage*
1 11 12 10 12 45
2 10 12 11 12 12 12 '12 12 93
3 12 11 12 12 11 12 12 12 94
4 11 11 12 11 12 12 11 12 92
5 10 12 12 11 45
6 12 11 12 12 47
Reversal Stage*
1 3 10 13
2 1 2 4 4 4 6 7 7 10 46
3 0 1 2 7 8 9 10 37
4 1 4 5 6 8 9 8 11 52
5 3 6 7 11 27
6 3 6 8 11 28

 

*12 trials per day

74

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR CONTROL GROUP OF EXPERIMENT III

 

 

 

Days
8's 1 2 3 4 5 6 7 8 9 Total
Acquisition Stage*
1 11 12 23
2 7 8 12 27
3 7 11 18
4 7 6 6 6 7 7 10 49
5 9 12 21
6 8 9 8 12 37
Reversal*
1 4 7 12 23
2 1 3 6 10 20
3 3 7 10 20
4 3 5 7 9 12 36
5 2 12 14
6 3 1 2 4 4 3 5 8 11 41

 

*12 trials per day

75

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR EXPERIMENTAL GROUP OF EXPERIMENT IV

 

 

 

Trials
S's 1-6 7-12 13-18 19-24 25 Total
Acquisition Stage*
1 3 6 6 6 21
2 4 5 5 5 19
3 3 4 5 6 18
4 3 4 4 5 4/4 20
5 4 4 6 5 19
6 2 4 5 5 16
Overtraining Stage**
S's 1 2 Days3 4
1 14/15 11/12 11/12 11/12 47/51
2 14/15 12/12 11/12 12/12 49/51
3 14/15 11/12 11/12 12/12 48/51
4 14/15 11/12 11/12 12/12 48/51
5 13/15 10/12 11/12 12/12 46/51
6 15/15 11/12 12/12 11/12 49/51
Beversal***
S's 1-5 6-10 Tr17£§5 16-20 21-25 26-30
1 O 4 4 4 5 4 21
2 1 2 4 5 4 5 21
3 1 4 4 4 4 4 21
4 o 3 3 5 3 5 19
5 1 3 4 4 5 5 22
6 0 1 3 4 5 5 18

 

*12 trials per day
**n/m, i.e., n correct Rs in m trials

***15 trials per day

76

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR CONTROL GROUP OF EXPERIMENT IV

 

 

Trials
S's 1-6 7-12 13-18 19-24 25-30 31-36 Total

 

Acquisition Stage*
1 5 6 23
20
22
19
4/4 22

O\\n-F‘K»JN
O\O\O\O\\h0\

5 4
3 6
5 5
4 4
4 4

mm O\\J'\ FT 0\

Reversal Stage**
Trials

1-5 6-10 11-15 16-20 21-25 26-30

5 3 5 5 5 5 23
24
25
27
21

H

1

mm PK») N
C) to p: *4

A) kn -F #- U1
U1 to \$ v\ \n
\n #' U1 kn .p
3' U1 U1 \n .p
\n \m U1 \n Km

21

* 12 trials per day
** 15 trials per day

77

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR SMALL REWARD EXPERIMENTAL GROUP OF EXPERIMENT V

 

 

Trials
S's 1-4 5-8 9-12 13-16 17-20 21-24 25-28 29-32 33- Total

 

Acquisition Stage*

1 3 4 3 3 2 3 4 3 1/1 27
2 2 2 3 4 2 4 17
3 3 2 3 3 3 4 18
4 2 2 3 3 3 3 4 20
5 4 2 3 3 3 4 19
Overtraining Stage *
1 4 3 3 _ 3 4 4 21
2 3 3 3 3 4 3 19
3 4 3 3 3 3 4 20
4 4 3 4 4 4 3 22
5 3 3 4 3 4 4 21
Reversal Stage**

Trials '
S's 1-5 6-10 11-15 16-20 21-25 26-30 31-35 36-40 41-45
1 3 3 5 4 4 4 23
2 2 3 3 3 4 5 21
3 1 4 4 5 3 4 21
4 2 2 3 4 4 4 19
5 2 3 4 4 4 3 4 5 5 34

 

* 12 trials per day
** 15 trials per day

‘
I

78

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR LARGE REWARD EXPERIMENTAL GROUP OF EXPERIMENT V

 

 

Trials
S's 1-4 5-8 9-12 13-16 17-20 21-24 Total
Acquisition Stage*
1 3 4 4 3 4 3 22
2 2 3 4 4 4 3 20
3 4 4 3 3 4 3 20
4 3 4 4 4 4 4 23
5 3 2 3 4 4 3 19
Overtraining Stage *
1 4 3 4 4 4 4 23
2 3 3 3 4 3 4 20
3 4 4 4 4 4 4 24
4 3 4 4 4 4 4 23
5 4 4 3 4 4 3 22
Reversal Stage**

Trials
S's 1-5 6-10 11-15 16-20 21-25 26-30
1 3 5 5 13
2 3 5 5 13
3 1 5 5 5 5 4 25
4 3 5 5 13
5 3 5 4 12

 

*12 trials per day
** 15 trials per day

79

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR SMALL REWARD CONTROL GROUP OF EXPERIMENT V

 

 

Trials Trials ** Total
S's 1-4 5-8 9-12 13-16 17-20 21 Total to Crit. Trials

 

Acquisition Stage*

1 2 3 4 9 4 12
2 3 2 4 9 5 12
3 2 2 3 3 1 1o 11 17
4 2 2 2 2 3 11 13 19
5 2 3 2 3 1 11 11 17
6 2 3 2 3 2 4 16 19 25

Reversal Stage***
Trials

S's 1-5 6-10 11-15 16-20 21-25 26-30

1 3 3 4 4 4 5 23
2 2 4 4 4 4 4 22
3 2 3 3 4 4 5 21
4 2 4 4 4 5 5 24
5 2 3 3 5 4 5 22
6 1 4 3 5 5 4 21

 

* all 6's run in one day until criterion reached
** not including criterion trials

*** 15 trials per day

80

TOTAL CORRECT FOR EACH DAY OF TESTING
FOR THE LARGE REWARD CONTROL GROUP OF EXPERIMENT V

 

 

 

S's 1-4 5-erZETZ 13-16 Total Egigiitf* $85833
Acquisition Stage*

1 2 3 2 4 11 9 16
2 2 3 4 9 4 12
3 2 3 4 9 3 12
4 3 4 3 10 o 12
5 2 3 3 2 10 7 14
6 2 3 3 2 10 7 14
Reversal Stage***

S's 1-5 6-10 11-15 763268 21-25 26-30 31-35

1 3 4 4 5 4 5 29
2 3 4 5 12
3 2 5 5 12
4 2 5 5 12
5 3 5 4 12
6 5 5 4 14

 

* all 6's run in one day until criterion reached
** not including criterion trials

44* 15 trials per day