THE EFFEC? 0F SPACENG O? TRAINENG AND

SURGICAL EXTIRPATION ON
CONDITIONENG IN EARTHWORMS

Thesis for We Degree of M. A.
MICHIEAN STATE UNIVERSITY
Kliem Ralph Miner

1958

UPrngmv“
5“ A5 5 55M“.

{55 BACK 05: (“On

 

     
 

LIBRARY

Michigan State
University

 

THE EFFECT OF SPACING OF TRAINING AND
SURGICAL EXTIRPATION ON CONDITIONING IN EARTHWORMS

By

Kliem Ralph Miller

A THESIS

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

MASTER OF ARTS

Department of Psychology

1958

ABSTRACT

This study was designed to investigate in a factorial
design, with earthworms, Lumbricug Terrestris, the effects
of (a) removing the ganglia (anterior five segments removed)
and (b) spacing of trials on (a) the conditioning of a with-
drawal response and (b) reconditioning of this response after
a rest period. The following experimental groups were used:
(1) normal spaced, (2) normal massed, (3) Operated massed,
and (4) Operated spaced.

Sixty earthworms were randomly assigned to groups:
normal spaced, NS; normal massed, NM; Operated spaced, OS;
Operated massed, OM; operated control spaced, 008; and Opera-
ted control massed, OCM. The forty Operated Ss had their
dorsal ganglia and anterior five segments removed while the
ganglia of the remaining 20 Ss were left intact. Massed con-
ditioning trials had a ten sec. inter-trial interval and
spaced conditioning trials had a 90 sec. inter-trial interval.

The apparatus consisted of a clear, plastic tube into
which the S was placed. The tube and a bell buzzer were
mounted on a board with a photo-flood light centered above
the tubes. Vibration of the bell buzzer provided the CS, and
light from the photo-flood provided the US. For groups NS,
NM, OS, and OM, the CS was presented for 6 sec., during the

last 2 sec. of which the US was presented. The occurrence

and latency of a withdrawal response during the first 4 sec.
of a vibration were recorded. Groups 003 and OCM received

6 sec. of vibration alone and the occurrence and latency of
a withdrawal response during the first 4 sec. were recorded.

Groups NS, NM, OS, and OM received 80 conditioning
trials followed by a 20 min. rest and ten post-rest condi-
tioning trials. Groups 008 and OCM received 80 stimulus
presentations followed by a 20 min. rest and ten post-rest
presentations.

The results clearly showed conditioning in groups NS,
NM, and OM on both measures of conditioning, while group
OS did not show conditioning. The performance Of control
groups, 008 and OCM, clearly showed the Operation did not
account for the conditioning found in groups OS and OM.
Group NS conditioned better than group NM on both measures
of conditioning. Group OM clearly showed conditioning while
group OS did not condition at all. Comparisons between the
Operated and normal groups showed that groups NM and OM do
not differ significantly, while group OM conditioned and
group OS did not condition at all. All groups showed a de-
crement in performance on the post-rest trials.

The results of this experiment showed that (a) spaced
trials facilitated conditioning fOr normal Ss; (b) removing
the "cerebral” ganglia did not affect conditioning with massed
trials; (0) removing the "cerebral" ganglia prevented condi-

tioning with spaced trials; (d) a rest period in the apparatus

lead to a post-rest decrement for all groups.

ACKNOWLEDGEMENTS

The writer is deeply indebted to Dr. Stanley C.
Ratner for his advice in formulating this experimental
problem and for his continued OOOperation in all matters
dealing with this thesis.

The writer is also grateful to Drs. M. Ray Denny and
A. M. Barch for their constructive criticism of the manu-

script.

TABLE OF CONTENTS

Page
INTRODUCTION . . . . . . . . . . . . . . . . . . . l
Pr0blem O O O O O O O O O I O C O O O O O O O 2
METHOD 0 O O O O O O O O O O O O O O O O O O O O O 5
subjeCts O 0 O O O O O O O O O O O O O O O O 5
Apparatus . . . . . . . . . . . . . . . . . . 3
Procedure 0 O O O O O O O O O O O O O O O O O 7

RESULTS 0 O O O O O O O t. O O O O O O O O O O O 0 lo

Conditioning . . . . . . . . . . . . . . . . lO
Sensitization Control for Conditioning . . . 17
Comparing Massed Spaced Groups . . . . . . . 18
Comparing Normal and Operated Groups . . . . l9
Post-rest Conditioning . . . . . . . . . . . 21

Relationship Between S's Length and Values of
UR'S O O O O O O O O O O O O O O O O O O O O 24

DISCUSSION 0 O O O O O O O O O O O O O O O O O O O 25

Conditioning . . . . . . . . . . . . . . . . 25
Comparing Massed and Spaced Groups . . . . . 27
Comparing Normal and Operated Groups . . . . 28
Post-rest Conditioning . . . . . . . . . . . 28

Relationship Between S's Length and
Total Number Of UR's . . . . . . . . . . . . 30

SUMMARY .

REFERENCES

APPENDIX A

APPENDIX B

Page
51

54

35

41

LIST OF TABLES

Table Page

1 A statistical analysis employing the Bino-
mial Test to determine if the increase or
decrease in S's performance during condition-
ing trials is due to chance alone. 36

2 A statistical analysis of the differences
between the Operated conditioning groups and
the Operated control groups, in order to de-
termine whether the Operation may have accoun-
ted for conditioning. The Krieskal-Wallis
one—way analysis of variance test was em-

ployed. 57

5 A statistical analysis Of the differences in
conditioning between massed and spaced groups
employing the Mann-Whitney U test. 58

4 A statistical analysis of the differences in
conditioning between normal and Operated
groups employing the Mann-Whitney U Test. 58

5 A statistical analysis employing the Bino—
mial Test to determine if the increase or
decrease in S's performances after rest is
due to chance alone. 59

6 A statistical analysis of the differences in
recovery after rest between massed and spaced
training groups employing the Mann-Whitney g
Test. 40

7 A statistical analysis of the relationship
between S's length and values of UR's, em-
ploying a rank order correlation coefficient. 40

Figure

LIST OF FIGURES

A photOgraph of the apparatus used in this
study.

A close-up photograph of an S in one of the
animal tubes.

Percentage of CR's for groups NS and NM.

Percentage of CR's for groups OS, OM, 008,
and OCM.

Median latency Of CR's for groups NS and NM.
Median latencies computed only from trials
in which CR's occurred.

Median latency of CR's for groups OS, CM,
008, and OCM. Median latencies computed
only from trials in which CR's occurred.

Median latency of CR's for groups NS, NM,
OS, and OM. Median latencies computed from
all trials for each block.

Page

12

13

l4

15

l6

I. INTRODUCTION

Until recently the investigation Of classical condi-
tioning in annelids has been largely neglected. The marine

annelid flydroides dianthus, received some attention from

 

Yerkes (10), who conditioned two specimens of Hydroides
dianthus to react negatively to photic radiation. In 1912,
Yerkes (11) found that a manure worm, Allolobophora foetida,
was capable of acquiring certain modes of reaction which in-
volved a definite direction of movement and the association

of two stimuli. Yerkes then removed the worm's anterior gang-
lia and after further experimentation with the animal conclu-
ded that performance of the worm in a T-maze was not dependent
on the ganglia. Heck (8, pp. 213) trained a number of Eisenia
foetida and Lumbricus terrestris and found that the removal of
the ganglia did not prevent the learning of a sample turning re-
Sponse. However, at present no information is available con-
cerning the effect of ganglia removal on annelid performance,
using a classical conditioning paradigm.

In 1957, Ratner and Miller (4) clearly demonstrated
classical conditioning of a withdrawal response in earth-
worms, Lumbricus terrestris, under massed training condi-
tions using a vibratory stimulus as the CS and light as the

US.

The fact that the mean per cent of CBS during the last
10 conditioning trials only reached 42.50 per cent may be
due to the massing of the conditioning trials. However,
past studies have not demonstrated a significant difference
in performance of 83 under conditions of massed and spaced

training with invertabrates.

Problem

The purpose of the present experiment is to investigate
in a.£actorial design, with earthworms, Lumbricus terrestris,
the effects of (a) removing the ganglia (anterior 5 segments
removed) and (b) spacing of trials on (a) the conditioning
Of a withdrawal response and (b) reconditioning of this re-
sponse after a rest period. For this purpose the following
experimental groups were used: (1) normal spaced, (2) nor-

mal massed, (3) Operated massed, and (4) Operated spaced.

II. METHOD

Subjects

Sixty earthworms, Lumbricus terrestris, varying in size
from 121 to 240 segments were studied. The worms were ob-
tained from a plot of ground on the Michigan State University
campus and were placed in the apparatus within one hour after
removal from their natural habitat. During this transfer
period Ss were kept in Sphagnum moss.

Forty of the Ss had their pharyngeal ganglia (anterior
5 segments removed) removed while the remaining 2O 83' ganglia

were left intact.

Apparatus
Two photographs of the apparatus are shown.in Fig. l

and 2 respectively. Two identical units were constructed
so that two 83 could be run at the same time. One S was
run during the inter-trial interval of the other. The main
part of the apparatus consisted of 2 clear, round, Koroseal
tubes vented on the tOp and sides with 1 mm. holes at 5 cm.
intervals. Each tube was 58 1/4 cm. long with an inside
diameter of 8 mm. Each tube was mounted on a separate ply—
wood base, with both ends fastened together by thin OOpper
wire, thereby producing a circular runway for the animal.

Each tube base was 24 cm. square with 6 l/2 cm. sides over

_ 3 -

 

 

 

A photograph showing the apparatus used in this
study.

 

 

 

Figure 2. A close-up photograph shmwing an S in one of the
two identical S housing tubes.

which a lid could be placed to shield one S from the uncon-
ditioned stimulus (US) while the other S was receiving a
trial. A 6 volt d.c. door bell, with the bell removed, was
attached securely on the base at the center of each subject
tube and was used as the conditioned stimulus (CS). Each
door bell was separately controlled by a toggle switch per-
mitting only a single CS source to be operated by the timer
at any one time. One #2 (G.E.) photo flood bulb in a 50.5
cm. aluminum reflector was mounted 45 cm. above and midway
between both subject tubes and constituted the source of
the US. An 8 candle power, dark red bulb positioned 1 cm.
below and at the outer edge of the US reflector, provided
general illumination.

The CS and US were timed and automatically presented
by three interval timers. Inter-trial-intervals were timed
with a stOp watch. CR latent periods were measured by a
timing clock which was activated by the CS timer and stOpped
by a micro switch Operated by the experimenter.

All parts Of the apparatus that could have produced
noise or vibration were padded with foam rubber 1 cm. thick.
This includes the stOp watch holder, subject enclosure, lid
and timing clock. To further control extraneous vibration,
all the equipment was placed on a 75 x 71 x 1 cm. plywood

base which was clamped securely to a fairly rigid table.

Procedure

88 in the Operated groups were placed on a disecting
board and the first 5 segments were cut off with a sharp
razor blade. Immediately following the Operation 88 were
again placed in sphagnum moss for a 10 min. interval before
being placed in the apparatus. At this time the previously
sectioned segments were examined to make certain the ganglia
were present. If the ganglia was not present, both 83 were
discarded and the above procedure repeated.

The tube which was to contain the S was moistened with
.6 cc. of water immediately prior to 8'5 entry. The S was
transported from the sphagnum moss by hand the the S's an-
terior portion was placed in one end of the tube permitting
the S to advance into the Koroseal tube. The ends of the
tube were then fastened together with thin OOpper wire in
such a way that the S had a complete circle to transverse.
The above procedure was followed for each S in each tube.

An adaptation period which allowed the S to become addp"
ed ' to the 8 candle power general illumination light started
immediately after both Ss were in their respective tubes,
and continued for 20 minutes preceding experimental study.

The 83 were randomly assigned to one of six groups con-
sisting of 10 $3 each: The normal subject massed training
group (NM), the normal subject spaced training group (NS),

operated . . .
theAsubject massed training group (OM), the Operated subject

spaced training group (OS), the Operated subject spaced
training control group (OCS), and the Operated subject massed
training control group'(OCM).

Group NM was given a total of 90 conditioning trials
with.awzp min. rest interval between trials 80 and 81. The
inter-trial-interval for this group was 10 sec. for each
trial. The vibratory stimulus (CS) was presented for 6 sec.,
after the first 4 sec. of vibration the light (US) was pre-
sented for 2 sec. Thus, the CS and US overlapped for 2 sec.
and were terminated together. A withdrawal response occur-
ring during the first 4 sec. of the CS, prior to the pre-
sentation of the US, was recorded as a conditioned with-
drawal response.

The unconditioned response (UR) to light in normal Ss
consisted of a rearing and withdrawal of the anterior seg-
ments of S's body. If 83 were moving at the time of stimu-
lation the response consisted of an abrupt stOp followed by
the above response. In Operated Ss the UR was either iden-
tical to the above or characterized by a forward movement
of the anterior segments. Thus, a response of either type
which occurred prior to the US was recorded as a conditioned
response (CR). The latent period which intervened between
the onset of the CS and the beginning of the CR was recorded.
This measure was introduced because it is generally a more

sensitive measure of response strength than frequency of CBS.

(3: PP0 528)

For group NS, trials were administered in the same
manner as for group NM, except that the inter-trial interval
was 90 sec.

Groups OM and OS differ respectively from groups NM
and NS in that Ss in groups OM and OS underwent the Opera-
tion procedure previously described.

Groups 008 and OCM were employed to determine the num-
ber of responses which might occur as a result of the opera-
tional procedure sensitizing the animals to the vibratory
stimulus. Ss in these groups were treated in the same man-
ner as groups OM and OS except the US was not presented at
any time. The 20 min. rest interval between trials 80 and

81 applied to all groups.

III. RESULTS

Conditioning

Two measures of conditioning were used for each group
during the entire procedure, viz: percentage of CR's and
response latency. Figure 5 presents the percentage of CR's
during conditioning for the normal groups, groups NM and NS,
for each block of ten trials and Fig. 4 shows this for the
Operated groups, groups DM and DS. To determine if condi-
tioning occurred within each group binomial tests (7, pp.
56-42) were made for each group by comparing the percentage
of CR's in block I (trials 1-10) with the percentage of CR's
in block VIII (trials 71-80), for each S. The binomial tests
yielded significant one—tailed probabilities for groups
NM (p = .001), NS (p = .01), OM (p = .05), and OS (p = .05).
That is, there was a significantly higher percentage of CR's
on the last block Of trials than on the first block for groups
NS, NM, and 0M. This was taken as evidence for conditioning.
Group OS had a significantly higher percentage of CR's on the
first block of trials than on the last block of trials. This
indicated a lack of conditioning. Table 1 in the appendix
presents details Of performance for each group.

Figures 15 and 16 show the median latencies of CR's for

the normal and Operated groups for each block of ten trials.

- 10 -

11

Due to the non-homogeneity of the latency scores through-
out this study, median latency scores were computed and used
in all the statistical analyses. Figure 7 shows the learning
curves for groups NS, NM, OS, and OM. The scores in Fig. 7
consist of total median latencies. That is, the median
latency scores were computed using the total number of trials,
therefore any block of 10 trials in which the total responses
did equal 50 per cent were given a median latency score of
4.0 secs. Binomial tests were again used to determine if con-
ditioning occurred for the four groups as evidenced by this
latency measure. The median latency for the first block of
trials (trials 1-10) was computed for each S and compared,
as stated before, with the median latency of the eighth block
of trials (trials 71-80). The binomial tests yielded signi-
ficant one-tailed probabilities for groups NM (p = .01),
NS (p = .01), OM (p = .05), and OS (p = .01). It can be seen
in Figs. 5 and 6 that groups NM, NS, and 0M showed decreases
in latency indicating conditioning, whereas group OS showed
an increase in latency of response, indicating the absence
of conditioning. Table l in the appendix presents additional
data for these groups.

In summary, the conditioning groups NS, NM, and 0M
clearly showed conditioning, on both measures of conditioning,
while group OS did not show conditioning on both measures of

conditioning.

12

°/oCR’8
A
O

 

 

 

0*- r—.——.—-

I255 4 5 e 7 saesrs
BLOCKS 0F TRIALS

Figure 5. Percentage of CR's for groups NS and NM.

80
7o
60

50

m
E:

o 40«
o\°

30-

, 20

IO

C)

 

15

 

 

525 4 5 S 7 8REsT§
BLOCKS OF TRIALS

Percentage of CR's for groups OS, 0M, OCS, and OCM.

.zi

l4

:5
. 9

 

01.9-91

I

I

I

5

3
Z
Z

4

IO N-pI OI
.94

'55

A

 

E5

'01

 

_MED|AN LATENCY OF CR’SISEC)

 

 

0 .

i7? 3 4 5 56—17 éRE'sre
BLOCKS OFTRIALS

Figure 5. Median latency of CR's for groups NS and NM.
Median latencies computed only from trials in
which CR's occurred.

15

. .5
9.91.?

A

(A5 OFNC R'C§(SEC8N

 

51
A

c).
E I
I— l .5‘
j . —— o s ‘
ZIOJ -—---- o M
s 4! ................ 0 CS
8 .5‘ -----.-- O C M
'2

 

 

1'2 3 is e 7 eaeér‘s
BLOCKS 0F TRIALS

0

Figure 6. Median latency of CR's for group OS, 0M, OCS, and
OCM. Median latencies computed only from trials
in which CR's occurred.

m s» .w .A
-52.? Q

'9‘

A

 

N
O

'01

t3

'9‘

 

MEDIAN LATENCYOF CR’SISEOJ

d

0

Figure 7.

 

0.0.... “ a
I ,i
‘. "u. “ .q

I ' I :I
. -. I .N
T in... I w
‘ fl... ‘ l
\ .
6., -.‘.
~ 0 ..O‘
\ .‘ .“‘ ..
0‘ t .,
. l
‘ O
0‘ I
0‘ I
5' I '.\
‘0' 0’. \.
‘0
\

-——-OS
--—----OM
~-«-~-~ N S
mm N M

I Eé 3 4 S 6 7 BRESTS
BLOCKS 0F TRIALS

Median latency of CR's for groups NS, NM, OS, and

OM. Median latencies computed from all trials
for each block.

1?

§§psitization Control for Conditioning

Since the Operation pgg_§g might have accounted for any
behavioral change shown by groups OS and 0M during condition-
ing trials, for example by sensitizing them to vibration,
two Operated control groups were run with vibration alone.
One group was run with massed presentations of vibration and
the other with spaced presentations. The data for these
groups, OCS, Operated control spaced, and OCM, Operated con-
trol massed, are shown in Figs. 4 and 6. It can be seen
that these groups showed almost no overall increase in per-
centages of responses and little change in overall response
latency after the first block of trials.

The Kruskal-Wallis one-way analysis of variance (7, pp.
184-195) was employed to test the over-all differences be-
tween groups, because Of non-homogeneity of variance as
tested by Bartlett's test (9, pp. 194-195). Group OS made
a median of 15 CR's, Group 0M a median of 51 CR's, Group
OCS made a median of 17 CR's and Group OCM a median of 11.5
CR's for trials 1-80. That is the Ss in group 0M made signi-
ficantly more CR's than $3 in groups OS, OCS, and OCM,

(H a 16.16 p4< .01). See also Table 2 in the appendix.

In terms of median latencies of CR's made during trials
1-80, group 0S made a median latency of 2.59 secs., group
0M made a median latency of 2.0 secs., group OCS made a

median latency of 2.07 secs., and group OCM made a median

18

latency of 2.66 secs. That is the Ss in groups 0M made
significantly shorter scores than 83 in groups OS, OCS, and

OCM, (H = 15.96 p <..Ol). See also Table 2 in the appendix.

Comparing Massed Spaced Groups

In order to investigate the effects of the spacing of
training on conditioning, the massed groups were compared
with the spaced groups at the end of learning in that this
was where the difference would be expected to occur. Mann-
Whitney U tests (4, pp. 116-127) were applied to the total
number Of CR's made by the Ss in each group for trials 71-80.
This test (henceforth referred to as U test) was employed
because of non-homogeneity of variance as tested by Bartlett's
test of homogeneity Of variance (9, pp. 194-195). Group NS
made a median of 8.85 CR's and group NM made a median of
6.70 CR's for trials 71-80. A U test of the difference be-
tween these groups yielded a significant value of U,

(U a 26, p <1.05). That is, the normal spaced group made
reliably more CR's than the normal massed group.

Group OS made a median of 1.5 CR's and group 0M made a
median of 6.5 CR's for trials 71-80. A U test of the dif-
ference between these groups yielded a significant value of
Q’(Q = 1.5, p <1.001), indicating the Operated spaced group
made significantly fewer CR's than the Operated massed group.

In terms of latencies, group NS had a median latency of

.84 secs. and group NM had a median latency of 1.55 secs. for

19

trials 71-80. A U test of the difference between these
groups yielded a significant value of U’(U a 20.5, p <1.05).
That is the normal spaced group made reliably shorter la-
tencies than the normal massed group.

Group OS had a median latency Of 5.12 secs. and groups
0M had a median latency of 1.58 secs. for trials 71-80. A
U test of the difference between these groups yielded a
significant value of Q, (g = 5, p <..001). That is, the
Operated spaced group made significantly longer latency
scores than the Operated massed group. Table 5 in the ap-
pendix presents additional information for the above groups.

In summary, the normal spaced group conditioned better
than the normal massed group on both measures of condition-
ing. The Operated massed group clearly showed conditioning

while the Operated spaced group did not condition at all.

Comparing Normal and Operated Groups

To evaluate the effects of the Operation on condition-
ing each Operated group was compared with its similarly
trained normal group. Mann-Whitney U tests (7, pp. 116-127)
were again used to evaluate the difference between the groups
using the data from trials 21-80. The first two blocks of
trials were omitted due to the previous finding (5) that the
CS, vibration, elicited some avoidance responses early in
conditioning and thus contaminated the data for the first

two blocks of trials.

20

The median of the total number of CR's for trials 21—80
for group NS was 45.5 and the median of total number of CR's
for trials 21-80 for group OS was 11. A U test of the dif-
ference between total values of CR's for groups NS and OS
yielded a significant value of U (U = 11, p‘( .001). That
is, group NS made reliably more CR's than group OS during
conditioning.

The median of the total number of CR's for trials 21-80
for group NM was 55.5 and the median Of the total value of
CR's for group 0M was 27 for trials 21-80. A U test of the
difference between total values of CR's for groups NM and 0M
yielded a non-significant value of g (Q_= 49.5, pj> .05).

The median of the latencies of CR's for trials 21-80
for group NS was 1.51 secs. and 2.48 secs. for group OS. A
U test of the difference between total values of CR's for
groups NS and 0S yielded a significant value of U (U = 8,
pI( .001). That is, group NS made reliably shorter latencies
than did group OS during conditioning.

The median of the latencies for trials 21-80 for group
NM was 1.86 secs. and 2.02 secs. for group 0M. A U test of
the difference between these groups yielded a non-significant
value of U (U = 52.5, p > .05).

In summary, comparisons between the Operated and normal
massed groups showed that the massed groups (NM and 0M) did

not differ significantly on either measure of conditioning.

21

However the difference was highly significant between the
Operated and normal spaced groups on both measures of con-
ditioning. That is, the normal spaced group was the best and

the Operated spaced group showed no conditioning.

Post-rest Conditioning

Data from trials 81-90 in Figs. 5 and 4 show the per-
centage of CR's after rest for the normal and Operated groups.
In all groups the percentage of CR's drOpped considerably af-
ter rest.

Statistical analysis of the post-rest conditioning data
were made by binomial tests comparing the percentage Of CR's
in block VIII (trials 71-80) before rest with the percentage
of CR's in block IX (trials 81-90) after rest. The binomial
tests yielded significant one-tailed probabilities for groups
NM (p = .001), NS (p = .001), 0M (p a .001), while a none
significant probability was Obtained for group 0S (p = .172).
That is, there was a significantly lower percentage of CR's
on the ninth block of trials than on the eighth for groups
NM, NS, and 0M. This was taken as evidence for a lack of
recovery after rest. Table 5 in the appendix presents addi-
tional information for these groups.

Data from trials 81-90 in Figs. 5 and 6 show the median
latency of CR's after rest for the normal and Operated groups.

In all groups the median of latency scores increased considerably

22

after rest. Due to the non-homogeneity of the latency scores,
median latency scores were computed and used in the statis-
tical analyses. Binomial tests were used to determine if
there were an effect of a recovery after rest. The median
latency for the eighth block of trials (trials 71-80) was
computed for each 8 and compared as stated before, with the
median latency of the ninth block of trials (trials 81-90).
The binomial yielded significant one-tailed probabilities
for groups NM (p = .001), NS (p = .001), 0M (p = .01), while
a non-significant probability was obtained for group OS

(p = .577). That is, there was a significantly higher
latency on the ninth block Of trials than on the eighth for
groups NM, NS and 0M. This was taken as evidence for a lack
of recovery after rest effect. Table 5 in the appendix
presents additional data for all groups.

In order to determine whether the spaced and massed
training groups differed in the amount of change after rest,
U tests were applied to the total number Of CR's made by Ss
in each group for the ninth block of trials (trials 81-90).
Group NS made a median of 6.5 CR's and group NM made a
median of 2.75 CR's for trials 81-90. A U test of the dif-
ference between these groups yielded a significant value of
g, (Q’s 15.5, p‘<..01). That is, the normal spaced group
made reliably more CR's than the normal massed group during

the recovery trials.

25

Group OS made a median of .5 CR's and 0M made a median
of 2.85 CR's for trials 81-90. A U test of the difference
between these groups yielded a significant value of U
(H =- 22, p = .025).

In terms of latencies, group NS had median latency of
1.14 secs. and group NM had a median latency of 2.94 secs.
for trials 81-90. A U test of the differences between
these groups yielded a significant value of U (U = 19.5,

p <L.025). That is, the normal spaced group made reliably
shorter latencies than the normal massed group.

Group 0S had a median latency of 1.5 secs. and group
0M had a median latency of 5.11 secs. for trials 81-90. A
U test of the difference between these groups yielded a non-
significant value of U (Q = 52, pj) .05). That is, the
Operated spaced group made shorter latency scores than the
Operated massed group, but the value of the difference was
not reliably different.

In summary, all the conditioning groups showed a decre-
ment in performance on the post-rest trials. This decrement
in 85 performance was significant in groups NS, NM, and 0M,
while the Ss in group OS did not show a significant decrement
in post-rest trials. Comparisons between the normal massed
and spaced groups showed the Ss in group NS performed signi-
ficantly higher than $3 in group NM on both measures of re-

covery. The comparisons between Operated massed and spaced

24

groups showed Ss in group 0M performed significantly higher
than $8 in group OS on the total number of CR's measure.
However, the difference between group OS and 0M was not
significant on the median latency of response measure of

recovery.

Relationship Between S's Length and Values of UR's

In order to investigate the relationship between the
S's length as measured by the number of S's segments and,
total number of UR's, rank order correlation coefficients
(7, pp. 202-215) were computed. The data for the massed and
spaced groups were combined.

Group NM made a median Of 70.5 UR's and Ss in this
group had a median length of 170 segments. Group NS made
a median of 64.5 UR's and Ss in this group had a median
length of 165.5. The correlation between lengths and number
of UR's was significant (rS = .55, P = .01). Also see Table
7 in the appendix.

Group 0M made a median of 45.5 UR's and Ss in this group
had a median of 167 segments. Group OS made a median of
51.5 UR's and Ss in this group had a median of 180.5 seg-
ments.5 The correlation between lengths and number of UR's
was significant (rS = .58, p = .05). See also Table 7 in

the appendix.

IV. DISCUSSION

Conditioning

Figures 5, 4, 5, 6, and 7 clearly indicate the presence
of conditioning in groups NM, NS, and OM, and an absence of
conditioning in groups OS. The statistical tests yielded
significant differences in Ss' performance from the first
block to the last block of conditioning trials using both
total CR's and median latency scores.

The data from the Operated control groups, OCS and OCM,
support the interpretation that the Operation pg; pg could
not have accounted for the increased avoidance responses in
group 0M. Groups OCS and OCM, tested only with the vibra—
tory stimulus showed lower performance levels than group 0M,
and did not differ from group OS, which did not condition.

The large number of responses and the short latency of
responses shown by all groups during the first and/or second
blocks of trials is interpreted as a reaction to the vibra-
tion (CS). Data from a previous study by Ratner and Miller
(5), using a prOper control group indicated that the vibra-
tory stimulus elicited UR's during the early part of experi-
mental trials. That is, neutral stimuli may produce responseS'
but not for long because of sensory adaptation (5, pp. 71-82).

Sensory adaptation, sometimes referred to as stimulus satiation,

_ 25 _

26

refers to the fact that with continued or repeated presenta—
tion of a stimulus, the stimulus loses its eliciting or
directing power.

Of particular interest was the lack of conditioning in
group OS. A recent classroom study (5) as discussed below
has essentially verified this finding of no conditioning if
Operated Ss are trained under spaced conditions. Three ten—
tative hypotheses may be advanced to account for the absence
of conditioning in group OS. First, the finding may be due
to a sampling error. The verification of the data found in
group OS by the classroom study mentioned above did not
support the sampling error hypothesis. The second hypothesis
is that time in the apparatus may have led to a biolOgical
deterioration of the Operated spaced group. That is, they
were in the S tube for approximately four hours while the
Operated massed 85 were in the apparatus for approximately
one hour. The Operation combined with the time in that ap-
paratus may have had deleterious effects. This hypothesis
was not supported by data Obtained from the classroom study
above (5) in which an operated group received massed training
after a two-hour rest period in the apparatus. This 0M group
conditioned readily.

The third hypothesis which may explain the lack of con-

ditioning in group OS would impute a "remembering or memory"

27

function to the pharyngeal ganglia. This hypothesis seems
relevant since the Operated massed group did show condition-
ing, which shows the Operated Ss are capable of learning con-
ditioned responses. That the highest level of conditioning
was present in group NS also seems to uphold the interpreta-
tion that the pharyngeal ganglia is necessary for spaced
conditioning. That is, we would expect group OS to condition
to a higher level than group 0M, since group NS conditioned
to a higher level than group NM.

An experimental design which might throw light on the
lack of conditioning in group OS would involve a long condi-
tioning procedure in which blocks of massed trials are inter—
spersed with blocks of spaced trials. According to the third
hypothesis above we would expect the massed group to condition

to a higher level than the spaced group.

Comparing Massed and Spaced Groups

Figures 5, 4, 5, and 6 clearly show a difference in per-
formance between the massed and spaced training groups for
both measures Of conditioning. That is, thespaced group, NS,
conditioned significantly better than the massed group, NM.
Group 0M conditioned significantly better than group OS, which
did not condition at all. Until the present study the massed-
spaced training effect had not been clearly demonstrated among
invertabrates. Yerkes (10) for example, found spacing did not

affect performance, using an instrumental paradigm.

28

However, the finding that group NS conditioned to a
higher degree than group NM is consistent with other find-
ings on the effects of massing vs. spacing in a condition-

ing situation (5, pp. 556-550).

Comparing Normal and Operated Groups

The lack of a significant difference between groups NM
and OM is consistent with findings by Yerkes (11), who trained
one worm Eisenia foetida in a T maze, and Heck (8, pp. 215),
who trained a number of Eisenia foetida and Lumbricus terres-
ppig. These workers found that removal of the ganglia did
not prevent the learning of a simple turning response. There
are no data in the literature which bear on the behavior of
the 0S group, which showed no conditioning at all. The hy-
potheses relating to the failure of the Operated-spaced

group to condition were presented earlier.

Post-rest Conditioning

Figures 5, 4, 5, 6, and 7 clearly indicate the absence
of any recovery after rest in groups NM, NS and 0M. That
is, as Opposed to recovery there was a significant decrement
in Ss' performances after rest. The statistical analyses
shows this decrement was greater than could be expected by
chance alone. Group OS did not show a significant increment

or decrement in performance since they did not condition.

29

The significant decrement found on post—rest trials in
groups NM, NS, and 0M was not what one would expect. That
is, typically massed practice followed by a rest leads to
improvement in performance and spaced practice followed by
a rest leads to little or no change in performance (5, pp.
556-550)-

The interpretation of the post-rest decrement is that
the rest period served as an extinction session. Since the
conditioning trials were presented at regular intervals for
all groups and since the rest took place in the conditioning
apparatus for all groups, it was possible that some temporal
conditioning took place which led to the occurrence of CR's
during the rest period in the absence of the US. Thus ex-
tinction was present.

This hypothesis may be tested by conditioning a group
of Ss and observing during a rest period the number of re—
sponses occurring during that interval of time in which the
CS previously occurred.

The graphs and statistical analyses Of post-rest per-
formance lend some support to this hypothesis. The Ss in
group NS performed significantly better than 83 in group
NM on both measures of post-rest conditioning.

Group NS gave a significantly higher post-rest perform-
ance than group NM and this finding is consistant with the
previous hypothesis accounting for_the lack of a recovery

after rest. That is, the massed group with an inter-trial

50

interval of 10 seconds experienced a greater number of ex-
tinction trials in the 20 minute rest-interval than the
spaced group. However, the difference between the groups
may simply be a function of the greater degree Of condi-
tioning which took place for group NS.

The comparisons between group OS and 0M show Ss in
group 0M performed significantly better than 83 in group
OS on the total number of CR's measure. However, the dif-
ference between OS and 0M was not significant on the median
latency of CR's measure of recovery. This finding was con-
sistent with the lack of conditioning in group 0S. That is,
operated 88 are not capable of conditioning under conditions

Of spaced training.

Relationship Between S's Length and Total Number Of UR's

The significant positive correlations found between S's
length (expressed in number of segments) and total responses
to the US is of interest in View of a previous finding by
Ratner and Miller (4) in which the only significant correla-
tion between S's length and total number of responses was
[found in a control group which received the US alone. This
correlation coefficient was negative. In the above study,
length was measured in metric units. Probably the most logi-
cal reason for the discrepency between the study mentioned
above and the present study was due to the more reliable meas-

ure of length used in the present study, namely counting the

S's segments.

V. SUMMARY

This study was designed to investigate in a factorial
design, with earthworms, Lumbricus Terrestris, the effects
of (a) removing the ganglia (anterior five segments removed)
and (b) spacing of trials on (a) the conditioning Of a with-
drawal response and (b) reconditioning of this response after
a rest period. The following experimental groups were used:
(1) normal spaced, (2) normal massed, (5) Operated massed,
and (4) Operated spaced.

Sixty earthworms were randomly assigned to groups:
normal Spaced, NS; normal massed, NM; Operated spaced, OS;
Operated massed, 0M; Operated control spaced, OCS; and Opera-
ted control massed, OCM. The fOrty Operated Ss had their
dorsal ganglia and anterior five segments removed while the
ganglia of the remaining 20 Ss were left intact. Massed con-
ditioning trials had a ten sec. inter—trial interval and
spaced conditioning trials had a 90 sec. inter-trial interval.

The apparatus consisted of a clear, plastic tube into
which the S was placed. The tube and a bell buzzer were
mounted on a board with a photo-flood light centered above
the tubes. Vibration of the bell buzzer provided the CS, and
light from the photo-flood provided the US. For groups NS,
NM, OS, and 0M, the CS was presented for 6 sec., during the

last 2 sec. of which the US was presented. The occurrence

- 51 -

52

and latency of a withdrawal response during the first 4 sec.
of a vibration were recorded. Groups OCS and OCM received
6 sec. of vibration alone and the occurrence and latency of
a withdrawal response during the first 4 sec. were recorded.

Groups NS, NM, OS, and 0M received 80 conditioning
trials followed by a 20 min. rest and ten post-rest condi-
tioning trials. Groups OCS and OCM received 80 stimulus
presentations followed by a 20 min. rest and ten post-rest
presentations.

The results clearly showed conditioning in groups NS,
NM, and 0M on both measures of conditioning, while group
OS did not show conditioning. The performance of control
groups, OCS and OCM, clearly showed the Operation did not
account for the conditioning found in groups OS and 0M.
Group NS conditioned better than group NM on both measures
of conditioning. Group 0M clearly showed conditioning while
group OS did not condition at all. Comparisons between the
Operated and normal groups showed that groups NM and 0M do
not differ significantly, while group 0M conditioned and
group OS did not condition at all. All groups showed a de-
crement in performance on the post-rest trials.

The results of this experiment showed that (a) spaced
trials facilitated conditioning for normal Ss; (b) removing
the "cerebral" ganglia did not affect conditioning with massed

trials; (0) removing the "cerebral" ganglia prevented

55

conditioning with spaced trials; (d) a rest period in the
apparatus lead to a post-rest decrement for all groups. A
positive correlation was found between Ss' length and values

of UR's.

1.
2.

10.

ll.

54
REFERENCES

Denny, M. R. (Personal communication).

Hess, W. N. Reactions to light in the earthworm, L.
Terrestris. g. Morphol. Physiol., 1924, 59:515—542.‘

Osgood, C. E. Method and theory ip experimental psychol-
351. Oxford, 1955.

 

Ratner, S. C., Miller, K. R. Classical conditioning in
earthworms, Lumbricus Terrestris. g. Comp. Physiol.

Psychol., (in press).
Ratner, S. C. (A classroom study).
Ratner, S. 0. (Personal communication).

Siegel, S. Nonparametric statistics for the behavioral
sciences. McGraw-Hill, 1956.

Warden, C. J., Jenkins, T. H., and Warner, L. H. Compara-
pive psychology, plants and invertebrates. Vol. II. New
York: Ronald, 1940.

Walker, H. M., Lev. J. Statistical inference. Henry Holt,

1955-

Yerkes, R. M. Modifiability of Hydroides dianthus V.
1. Comp. Neurol., 1906, 441-450.

Yerkes, R. M. The intelligence of earthworms, g. Anim.
Behav., 1912, 2:552-552.

"I

’I

APPENDIX A

35

56

TABLE 1

A statistical analysis employing the Binomial Test to de-
termine if the increase or decrease in S's performance dur-
ing conditioning trials is due to chance alone.

No. of S‘s No. of S's
showing showing

poorer per- better per- .
Group Scores Trials formance on formance on Cogiigince

trials 81-90 trials 81-90

than on 1—10 than on 1-10

 

NS CR's 1-10 1 9 p = .01
71-80

NS Latency 1-10 1 9 p = .01
71-80

NM CR's l-lO 0 10 p = .001
71-80

NM Latency 1-10 2 8 p = .05
71-80

OS CR's 1—10 8 2 p = .05
71-80

OS Latency 1-10 9 l p = .05
71-80

GM CR's 1-10 2 8 p = .05
71-80

GM Latency 1-10 2 8 p = .05

71-80

 

57

TABLE 2

A statistical analysis of the differences between the Operated
conditioning groups and the Operated control groups, in order
to determine whether the Operation may have accounted for con-
ditioning. The Krttskal-Wallis one-way analysis of variance
test was employed.

 

 

 

Group Scores Trials Rank Test 0525553256
0M R1 540.5
008 32 189.5
CS BB 175.5
OCM CR's 1-80 R4 89.5 H = 16.16 p.< .01
0M R1 255.0
005 R2 244.0
OS R5 191.0

OCM Latency 1-80 152.0 H = 15.96 p‘<..Ol

go

 

58
TABLE 3

A statistical analysis of thedifferences in conditioning
between massed and spaced groups employing the Mann-Whitney
U Test.

 

 

. Sum of Confidence

Group Scores Trials the ranks Test level

NS R1 129.0

NM CR's 71-80 R2 81.0 U = 26.0 p <..O5

NS R1 154-5

NM Latency 71-80 R2 75.5 E = 20.5 p < .05

0M CR's 71-80 R2 155.5 g = 1.5 p < .001

0M Latency 71-80 R2 152.0 E = 5 p < .001

TABLE 4

A statistical analysis of the differences in conditioning
between normal and Operated groups employing the Mann-Whitney
U Test.

 

 

 

Sum of Confidence
Group Scores Trials the ranks Test level
NS R1 144.0
OS CR's 21-80 R2 66.0 U = 11.0 p < .001
NS R1 147.0
OS Latency 21-80 R2 65.0 E = 8.0 p 4 .001
GM CR's 21-80 R2 104.5 U = 49.5 P > .05
NM R1 102.5
0M Latency 21-80 R2 86.5 H = 52.5 p > .05

 

39
TABLE 5

A Statistical analysis employing the Binomial Test to deter-
mine if the increase or decrease in S's performances after
rest is due to chance alone.

 

 

No. of S's No. of S's

showing showing
. - poorer per- better per- Confidence
Group Scores Trials formance on formance on level

Trials 81-90 Trials 81-90
than on 71-80 than on 71—80

 

NS CR's 71-80 10 0 p = .001
71-90

NS Latency 71-80 10 0 p = .001
71-90

NM CR's 71-80 10 O p = .001
71-90

NM Latency 71-80 10 O p = .001
71-90

OS CR's 71-80 7 5 P = .172
71-90

0S Latency 71-80 6 4 p = .577
71-90

OM CR's 71-80 10 0 p = .001
71-90 ‘ .

OM Latency 71-80 9 l p = .01

71-90

 

40

TABLE 6

A statistical analysis of the differences in recovery after
rest between massed and spaced training groups employing the
Mann-Whitney U Test.

 

 

 

Group Scores Trials Sum of Test Confidence
ranks level
NS CR‘s 81-90 R1 159.5 E = 15.5 p 4: .01
NM R2 70-5
NS Latency 81-90 R1 155.5 g = 19.5 p < .025
OS CR's 81-90 R1 77.0 U = 22 p < .025
os Latency 81-90 5R1 87.0 I_J_ = 52 p > .05
0M R2 125.0
TABLE 7

A statistical analysis of the relationship between S's length
and values of UR's, employing a rank order correlation coeffi-
cient.

 

 

 

Group Scores Trials D2 Test Coigigince
0M x 08 CR's 1-80 822.00 = .58 p = .05

 

APPENDIX B

"Raw data in folder."

41

 

5
5
5
5
ll
5
5
ll
5
5
5

IHIHHNHHI

83