OVERDUE FINES ARE 25¢ PER DAY
PER ITEM

Return to book drop to remove
this checkout from your record.

 

5......) (531/

 

 

EFFECTS OF STIMULUS RELEVANCE ON
MEMORY IN THE PIGEON

By
Rodney Charles Howard

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Psychology
1979

ABSTRACT

EFFECTS OF STIMULUS RELEVANCE ON
MEMORY IN THE PIGEON

By
Rodney Charles Howard

The effect of stimulus relevance on retention in the
pigeon was investigated in three experiments. In the basic
procedure, the occurrence of a 3 sec pre-stimulus indicated
that a forthcoming 15 sec white stimulus would be reinforced
on a variable interval schedule. The absence of a pre-
stimulus indicated that extinction was in effect during white.
In testing, retention functions were obtained by manipulating
the interval between the pre-stimulus and a 15 sec white
test stimulus. In Experiment 1, the effect of direct rele-
vance on retention was investigated by comparing the pigeon's
retention for a food pre-stimulus with that of a keylight
pre-stimulus. Results showed food retention to be superior
to keylight retention. In Experiment 2, the effect of
associative relevance was investigated by comparing the
pigeon's retention for two keylights after one received
direct food pairing. Results showed that retention for the
keylight paired with food was superior to the keylight lacking
such pairing. In Experiment 3, discriminative relevance was

investigated by manipulating the probability with which a

keylight pre-stimulus predicted a white stimulus. Results
showed that retention was a direct function of the probabil-

ity with which the keylight predicted the white stimulus.

To my wife, Sharleen,

for her valuable advice and consistent support

ii

ACKNOWLEDGMENTS

I would like to thank Dr. Mark Rilling for his aid and
supervision of my thesis research and for his assistance
in preparation of my thesis manuscript.

I would also like to thank the other members of my
thesis committee, Drs. M. Ray Denny, Glenn I. Hatton, and
Dozier W. Thornton for their constructive comments and their

valuable suggestions.

iii

TABLE OF CONTENTS

Page

LIST OF FIGURES . . . . . . . . . . . . . . . . . . v

INTRODUCTION . . . . . . . . . . . . . . . . . . . l
EXPERIMENT l

The Pigeon's Memory for Food Versus Wavelength 3

Method 4

Subjects 4

Apparatus 4

Procedure 4

Results and Discussion 8
EXPERIMENT 2

Retention of a CS . . . . . . . . . . . . . . 13

Method . . . . . . . . . . . . . . . . . 14

Subjects and Apparatus . . . . . . . 14

Procedure . . . . . . . . . . . . . 14

Results and Discussion . . . . . . . . . 20
EXPERIMENT 3

Retention of Discriminative Stimuli . . . . . 22

Method . . . . . . . . . . . . . . . . . 23

Subjects and Apparatus . . . . . . . 23

Procedure . . . . . . . . . . . . . 23

Results and Discussion . . . . . . . . . 26

GENERAL DISCUSSION . . . . . . . . . . . . . . . . 28

LIST OF REFERENCES . . . . . . . . . . . . . . . . 33

iv

Figure

LIST OF FIGURES

A diagram of the procedure used during
training. For an S+ trial either food
or green keylight preceded white. For
an 8- trial no stimulus preceded white

The average rate of responding for the
food pre-stimulus and the green key-
light pre-stimulus for the last three
days of training (X) and across reten-
tion intervals during testing

A diagram of the procedure used during
Experiment 2. The CS pre-stimulus
predicted the occurrence of 8+ and was
additionally followed by food on 75%
of the trials. The DS prestimulus
only predicted 8+ and was never
directly followed by food

The average rate of responding for the
DS, the CS alone, and the CS followed
by food for the last three days of
training (X) and across retention
intervals during testing .

The average rate of responding for the
SI pre-stimulus, the 82 pre-stimulus,
and the S3 pre-stimulus for the last
three days of training (X) and across
retention intervals during testing .

Page

. 10

l7

19

25

INTRODUCTION

In the recent surge of research on animal memory,
attempts have been made to characterize the structure of
memory in animals (Honig, 1978; Lett, 1975; Olton, 1978;
Revusky, 1971; Spear, 1973; Wagner, 1978). Most researchers
have generally assumed that animal memory consists of a long
term retention structure, a short term retention structure
and specific mechanisms for interaction between the two.
Evidence that animals can store information over extended
periods of time and then retrieve that information to
facilitate learning has come from delayed alternation
(Capaldi, 1971) and delayed reinforcement (Lett, 1973)
studies. There is also support for the view that information
can be actively maintained over short periods of time (Grant &
Roberts, 1976; Maki, Moe & Bierley, 1977) and that this pro-
cess, called rehearsal, may be necessary for the formation
of associations (Terry & Wagner, 1975; Wagner, Rudy &
Whitlow, 1973). Finally, the term "working memory" has been
coined to describe retention when information is available
in memory during a discriminative trial and then is terminated
(Honig, 1978).

While research abounds on memory mechanisms, little
attention has been directed toward the nature of information

stored in memory. There is some evidence, for example, that

l

2
a memory for food is important for learning. In runway
studies, rats have been able to learn that only alternate
trials are reinforced (Capaldi, 1967; Cogan & Capaldi, 1961)
or that magnitude of reinforcement changes across trials
(Capaldi & Cogan, 1967; Crespi, 1942). With the exception of
a few studies (Honig, 1978; Maki, Moe & Burley, 1977),
there is a paucity of research directly investigating memory
for stimuli. It seems likely that retention of a stimulus
would be affected by: (a) how directly relevant the stimulus
is to the animal (e.g., food to a hungry pigeon); (b) whether
the stimulus has relevant associations for the animal (e.g.,
a CS which predicts food); and (c) to what extent the stim-
ulus provides information about behavior associated with an
upcoming event (i.e., that future keypecking will be rein-
forced). The purpose of the present research was to
investigate memory for a stimulus as a function of the type

of relevance it has for the animal.

EXPERIMENT 1

 

The Pigeon's Memory for Food Versus Wavelength

One aspect which may affect retention is the amount of
direct relevance the stimulus has for the animal. Probably
the most relevant stimulus used in animal research is food.
Staddon (1974) has suggested that a pigeon's memory for food
is superior to its memory for other events. While some
evidence exists that animals show better retention for pre-
ferred reinforcers (Cowles & Nissen, 1937) and that the
occurrence of reinforcement contributes to the retention of
a pre-stimulus in a conditional discrimination task (Honig,
1978), direct evidence for better retention of food is sparse.

In the present study, retention of food was directly
compared with the retention of a keylight. If, as has been
suggested, food is a better remembered stimulus, its reten-
tion by the pigeon should be superior to the keylight. In
the procedure used (Reynolds & Catania, 1962; Weisman, 1976)
keypecking during a white keylight was differentially rein-
forced depending upon a discriminative pre-stimulus. If
either access to grain or illumination of the keylight with
a wavelength occurred, keypecking during white was reinforced.
In the absence of either event, extinction was in effect.
After the pigeon developed a discrimination between rein-
forced (S+) and non-reinforced (8-) trials, retention

3

4
functions were obtained by varying the delay between the
discriminative stimuli and the trial onset. If access to
food is retained in memory better than the presentation of
a keylight, responding should occur during white after
longer retention intervals when the discriminative stimulus

was food than when it was the keylight.

METHOD

' Subjects

Four naive White Carneaux pigeons maintained at 80%
of their free-feeding weights were used. The pigeons were

provided free access to grit and water.

Apparatus

 

A three key Lehigh Valley pigeon chamber was used. The
center key could be illuminated with either a green, red,
yellow or white stimulus using an Industrial Electronics
Engineers inline projector (Model #10-0W78-1820L). A house-
light consisting of a GE #1820 lamp remained on during the
session. A GE #1820 lamp illuminated the food hopper during
reinforcement. A fan which provided masking noise and

ventilation remained on during the session.

Procedure

 

Pigeons were initially magazine trained and auto-shaped
to peck the center key. An autoshaping trial, presented a

variable interval 30 sec (VI 30 sec) schedule, consisted of

5

the illumination of the key with a white stimulus for 8 sec
followed by access to a grain filled hopper for 3.5 sec.

Following autoshaping, pigeons received discrimination
training in the following manner. The center key was
illuminated with the white light for both reinforced (8+)
and non-reinforced (S-) trials. A reinforced trial began
with a pre-stimulus of either a dark keylight and 3 sec
access to grain or 3 sec keylight illumination with the green
stimulus. Following either event, the key was illuminated
with the white light (8+) and keypecking was reinforced on
a VI schedule. Only one reinforcement was collected per 8+
and the trial ended with reinforcement. The average duration
of the variable interval schedule was 15 sec (VI 15 sec).
For an extinction trial, no stimulus preceded the illumina-
tion of the key with the white stimulus (8-). Keypecking
during 8- was not reinforced and the duration of the trial
was 15 sec. Trials were presented in a mixed order with
all trials separated by a 30 sec intertrial interval (ITI).
A session consisted of 40 8+ and 40 8- trials with 8+
trials equally divided between those beginning with the food
pre-stimulus and those beginning with the keylight pre-
stimulus. Training was completed when a discrimination index
of .90 was reached for two consecutive sessions. The
formula is: gifgig: ‘where 8+ is the total number of
responses to the white (8+) stimulus, and S- is the total
number of responses to the white (S-) stimulus in a session.

Figure 1 shows a diagram of the training procedure.

Figure 1.

A diagram of the procedure used during training.
For an 8+ trial either food or green keylight
preceded white. For an 8- trial no stimulus
preceded white.

 

8+ TRIAL ‘

 

 

 

 

 

FOODQB GRN‘I I
3$EC
WH|fiE—J 15 SEC X
5" TRIAL
WH’T§* [ |§|TI 33$EC1
'5 SEC

 

TIME

‘V

8

Testing consisted of presenting probe delay trials
intermixed with normal training trials. In a testing trial,
either food or the keylight was followed by a delay interval
of 0 sec, 2 sec, 4 sec, 8 sec or 12 sec. During these probe
trials, keypecking to the white stimulus was not reinforced
and the stimulus remained on for 15 sec. Both pre~stimuli
were tested at each delay interval per session. A testing
session consisted of the 40 8+ and 40 8- trials, as in
training, with the 14 probe trials inserted between training
trials in a mixed order. All trials were separated by a
30 sec ITI. Each pigeon received seven sessions of testing
with the order of delay intervals counterbalanced across

sessions and pigeons.
RESULTS AND DISCUSSION

The left panel of Figure 2 shows the average rate of
responding for 8+ when preceded by food (filled circles) or
green (open circles) and for 8- (filled triangles) when
averaged over the last three days of acquisition training.
The rate of responding did not differ significantly as a
function of the discriminative stimulus used over this
period, E (3) = .60, p < .10.

The right panel of Figure 2 shows that rate of responding
during a white keylight preceded by either food or green
as a function of the interpolated retention interval. The
figure shows that responding during white dropped off at a

shorter delay interval with the green keylight than with food.

Figure 2.

The average rate of responding for the food pre-
stimulus and the green keylight pre-stimulus for

the last three days of training (X) and across
retention intervals during testing.

 

 

10

RESPONSES PER MINUT E

:on
.8...
mo...
8...
«or

mo
00

3..

 

uOI
NO!
.9...

 

 

I WOOD

OIIO xm<COI4
r mu

 

 

mm...mz....02 .24mm<>r Ammo.

11

A within-subject analysis of variance performed on the,
data showed that pre-stimulus and retention interval signifi-
cantly interacted to affect responding during white, F (6,8) =
18, p < .001. A test for simple main effects showed that
responding differed significantly at 2 sec, F (1,18) = 6.8,
p < .025, 4 sec, F (1,18) = 12, p < .005, 6 sec, F (1,18) =
7.98, p < .01, and 8 sec, F (1,18) = 6.95, p < .025.
Responding did not differ significantly at 0 sec, F (1,18) =
.16, p > .10, 10 sec, F (1,18) = .55, p > .10 and 12 sec,
F (1,18) = .51, p > .10.

The results indicate that pigeons were able to show
discriminative behavior over greater retention intervals
when food was the pre-stimulus. The differences in reten-
tion were obtained following training to near asymptotic
performance levels for both the food and keylight stimuli.
In testing, similar performance was obtained for both
stimuli at the 0 sec retention interval. It is unlikely,
therefore, that differences in responding observed at later
retention intervals resulted from inequities between stimuli
in the level of original learning. If the association
between the green keylight and white (8+) was not well
formed, some error in discriminative behavior would be
expected at the 0 retention interval resulting in an average
response rate for green below that obtained with food. In
general, the results support the notion that food produces
a more lasting memory than a keylight. Other interpretations,
however, are also possible. First, stimulus aftereffects

may result from food ingestion. These aftereffects may

12
persist long enough to effectively mediate the retention
interval through, perhaps, lengthening a peripheral sensory
trace (Sperling, 1963) for food and thereby retarding trace
decay. Roberts and Grant (1974) have shown that retention in
matching to sample is improved following increases in sample
stimulus duration. Another possibility is that separate
sensory traces were found for each aspect of food presentation.
Assuming variance in the rate of trace decay for any one
stimulus, the probability of a stimulus trace at the end of
the retention interval would be greater for the food pre-
stimulus, consisting of an auditory solenoid click, a hopper
light and grain presentation than the green pre-stimulus
consisting merely of keylight illumination. Therefore,
although retention of food was superior to a keylight, it
is equivocal whether this reflects differences in memory

for the events or some other factor.

EXPERIMENT 2

 

Retention of a CS

In Experiment 1, the superior retention of food over the
keylight supported the view that direct relevance affects
memory. A second factor which may affect memory is the
associative relevance of a stimulus. The most obvious member
of this category of stimuli is a C8. Although it is well
known that a C8 comes to control the behavior of an animal,
it is not known if an animal's memory for a stimulus changes
when it becomes a C8. Support has been gathered (reviewed
by Rescorla, 1978) for cued retrieval of an internal
representation of a US. It seems likely that if a stimulus
such as food forms a powerful memory as Experiment 1 suggests,
a stimulus which retrieves that memory should show some gain
in retention. To test this, a keylight pre-stimulus in the
present procedure was paired with food while another was not.
If associative relevance facilitates memory, the CS keylight
should show superior retention over the other keylights.
Also, while Experiment 1 was confounded because of problems
with the stimulus complexity of food presentation, this is
avoided when stimulus associations are manipulated because

physical properties can be held constant.

13

14

Some additional changes were made in the basic procedure.
To decrease the possibility that pigeons would discriminate
the varying duration 8+ from the fixed duration testing trial,
the 8+ trial was set at 15 sec during which a VI 15 sec
schedule of reinforcement was in effect. Although it is
possible that the pigeon could adopt a strategy in which a
slow rate of responding was maintained during any trial until
reinforcement indicated that the trial was an 8+, pilot work
indicated that such a strategy was not adopted. Also, to
decrease the difference between a testing trial in which a
delay was interpolated between the discriminative stimulus
and the test trial and an 8+ trial in which no delay occurred,
a 2 sec delay was interpolated between the discriminative

stimulus and the 8+ during acquisition.
METHOD

Subjects and Apparatus
Eight naive, White Carneaux pigeons, maintained at 80%
of their free—feeding weights were used. The apparatus was

the same as described in Experiment 1.

Procedure

 

The procedure followed the basic pattern described in
Experiment 1 with some important exceptions. An 8+ trial
began with the presentation of a discriminative pre-stimulus
on the key for 3 sec. After a 2 sec delay, the white

stimulus indicating an 8+ trial was presented. The duration

15
of the white (8+) stimulus was set at 15 sec, matching the
duration of the white (8-) stimulus. Keypecking was rein-
forced on a VI 15 sec schedule with the number of reinforce-
ments per trial ranging from 0 to 2. Two discriminative pre-
stimuli, red and green keylights, were used. One keylight
(DS) was never directly paired with food but consistently pre-
dicted the white (8+) stimulus. The other keylight (C8)
also consistently predicted the 8+ but was additionally
followed by a 3 sec access to grain on 75% of the trials.
Figure 3 shows a diagram of the procedure used. A session
consisted of 40 8+ and 40 8- trials. For half of the
pigeons, 20 8+ trials began with green, 15 began with red
followed by food and 5 began with red alone. For the
.remaining pigeons green was paired with food. Presentations
of the C8 followed by food insured conditioning, while
presentation of the C8 alone was intended to reduce the
possibility of a generalized decrement problem during
testing. All other aspects of training were the same as in
Experiment 1. Training ended when a pigeon obtained a .90
discrimination index for two consecutive sessions.

Testing was basically the same as in Experiment 1.
Retention functions were obtained for the D8, the C8 alone,
and the C8 followed by food. Each stimulus was tested at
2 sec, 4 sec, 6 sec, 8 sec, 12 sec, and 16 sec retention
intervals per session. The intervals were selected based on
pilot research. Testing ended when a pigeon received 6

sessions. Order of testing was controlled as in Experiment 1.

16

Figure 3. A diagram of the procedure used during
Experiment 2. The C8 pre-stimulus predicted the
occurrence of 8+ and was additionally followed
by food on 75% of the trials. The D8 pre-

stimulus only predicted 8+ and was never directly
followed by food.

17

8+ TRIAL

DISCRIMINATIVE STIMULUS (DSI

 

IOO% SSECj 05 !

I5 SEC; WHITE (5+) '310'SEC

CONDITIONED STIMULUS (CS)

 

 

 

 

 

 

 

 

 

 

 

 

75% + cs
m
as, TE 5
~ WHI +
15 SEC ( ) 3035c
25% I asec
I5 Steer WHITE (3+) 1 III
30 SEC
8— TRIAL
l WHITE Is-) I ITI
30 SEC

TIME

18

Figure 4. The average rate of responding for the D8, the C8
alone, and the C8 followed by food for the last

three days of training (X) and across retention
intervals during testing.

19

RESPONSES PER MINUTE

50
.30
.uo
N

NuaumqmmEZT
ooooooooooo
IIH

03

[j

l

I

I

I

 

 

II... On...
I Om ZIOZM
OI! OIIO Owl—«000

  

_ _._ . _ u

 

N b m m .N .m
mm...mz.2oz .z._.m2<>r A mmn .

20
RESULTS AND DISCUSSION

The left panel of Figure 4 shows the average rate of
responding for 8+ over the last 3 days of training when 8+
was signalled by the D8 keylight (filled squares) and by the
CS keylight alone (filled circles) or followed by food
(open circles). Responding during the 8- is also shown
(filled triangles). An analysis of variance performed on the
acquisition data shows that the rate of responding among the
stimuli did not differ significantly over the last 3 days
of training, is < 1.0.

The right panel of Figure 4 shows the rate of responding
during a white keylight preceded by either the D8 keylight,
the C8 keylight alone, or the CS keylight followed by food
as a function of the interpolated retention interval. The
figure shows that responding to the D8 keylight decreased at
a faster rate than responding to either the C8 keylight alone
or followed by food.

A within-subjects analysis of variance, performed on the
data, shows responding to differ significantly as a function
of the pre-stimulus used, F (2,14) = 9.12, p < .005, and as
a function of the retention interval tested, F (3,35) =
39.2, p < .001. A Duncan's multiple range test showed that
the C8 keylight did not differ as a function of whether or not
it was followed by food, but did differ significantly from the
D8 keylight, F (2,14) = 9.12, p < .005. A test for simple
main effects comparing the C8 and D8 keylights showed that

responding differed significantly at 4 sec, F (1,40) =

21
10.6, p < .005; 6 sec, F (1,40) = 17.6, p < .001; 8 sec,
F (1,40) = 15.8, p < .001; and 12 sec, F (1,40) = 7.4,
p < .01. Responding did not differ significantly at 0 sec
and 16 sec, Es < 1.0.

The results of this experiment show that retention of a
stimulus can be improved by its direct association with food.
There can be little doubt that good associations were formed
for both stimuli because discriminative behavior fOr each
keylight was excellent. The keylight associated with food,
however, showed better retention. In Experiment 1 a non-
memory hypothesis, involving sensory traces, was capable of
accounting for differences in retention. This was possible
because aCcess to food and the illumination of a response key
differed along a number of physical dimensions. In the
present experiment, physical aspects of the discriminative
stimuli, both keylights of approximately equal intensity,
were controlled. The superior retention for the keylight
associated with food, therefore, implicates memory processes.
Perhaps the superior retention for directly relevant and
associately relevant stimuli results from the same factor,
superior retention for memories of directly relevant
stimuli. Considerable evidence has been accrued for internal
representation of USS (Konorski, 1948; Rescorla & Heth, 1975)
and for C8 retrieval of these representations (Rescorla &
Cunningham, 1977; Terry & Wagner, 1975). In the present
experiment, the C8 pre-stimulus could retrieve an internal
representation of food which then mediates the retention

interval.

EXPERIMENT 3

 

Retention of Discriminative Stimuli

The results of the previous two experiments indicate
that the direct and associative relevance of a stimulus
affects its retention. In each study, retention was compared
against a stimulus which consistently predicted an upcoming
8+. This stimulus is considered to have discriminative
relevance. This occurs when a stimulus indicates that a
designated response will lead to a specified outcome. The
present procedure is ideal for investigating a pigeon's
memory for a stimulus with discriminative relevance for it
temporally separates the pre-stimulus from the required
response. To investigate discriminative relevance in
Experiment 3, the probability with which a stimulus pre—
dicted 8+ was manipulated. One stimulus (81) predicted 8+
90% of the time, a second stimulus (82) 50% of the time, and
a third stimulus (83) 10% of the time. For the remaining
trials, the stimuli were followed by an ITI. Previous
research has demonstrated that the amount of associative
strength which accrues to a stimulus is affected by the
probability with which it predicts reinforcement relative to
other stimuli (Wagner, 1969). While each pre-stimulus in

the present study is the best predictor of the 8+ on its

22

23
given trial, and discriminative performance is near asymp-
tote, the stimuli differ as to the consistency with which
they predict reinforcement. If this affects retention of a
stimulus, then the pre-stimulus which always leads to 8+
(81) should be superior to the stimulus which leads to 8+
50% of the time (82) which, in turn, should be superior to

the poorest predictor of 8+ (83).

METHOD

Subjects and Apparatus

 

Six naive White Carneaux pigeons maintained at 80%
of their free-feeding weights were used. The apparatus was

the same as described in Experiment 1.

Procedure

 

The pigeons were initially trained to peck the key using
autoshaping as described in Experiment 1.

Discrimination training utilized the same basic pro-
cedure as used in Experiment 2. The center key was illumin-
ated with the white light for both reinforced and non-
reinforced trials. Both 8+ and 8- trials were 15 sec in
duration and responding during the 8+ trial was reinforced
on a VI 15 sec schedule.

Three discriminative pre-stimuli, varying in the proba-
bility with which they predicted 8+ were used. Pre-stimulus
81 was followed, after a 2 sec delay, by white (8+) 90% of

the time, pre-stimulus 82 by white 50% of the time, and

24

Figure 5. The average rate of responding for the 81 pre-

stimulus, the 82 pre-stimulus, and the S3 pre-
stimulus for the last three days of training (X)
and across retention intervals during testing.

 

RESPONSES PER MINUTE

 

 

        

 

.Nofi.

20.. I m. woe.

00.. m oIIo m» mos

0|-.. mu 6.x.

co: > m-

8..

3..

mo:

«.0

so

8 -/ WW

no {/I-\. /

.o > .

o, F _ r t _ _ _ t
x 0 ~ 8 m m .o _m

mm...mz....02 .24mm<>r A mmo .

26

pre-stimulus 83 by white 10% of the time. When a pre-stimulus
was not followed by the white (8+) it was followed by the
standard 30 sec ITI. Keylight colors (red, green, yellow)
were counterbalanced across pigeons. A session consisted of
20 presentations of each discriminative stimulus for a total
of 60 trials. The 8- was also presented 60 times. Training
ended when a pigeon obtained a .90 discrimination index for
2 consecutive sessions.

Testing was basically the same as in Experiments 1 and
2. All three discriminative stimuli were tested at each
retention interval per session. The retention intervals
were 2 sec, 4 sec, 6 sec, 8 sec, 10 sec, and 12 sec. Testing
ended when a pigeon had received 6 sessions. Order of testing

was controlled as described in Experiment 1.

RESULTS AND DISCUSSION

The left panel of Figure 5 shows the average rate of
responding for 8+ over the last three days of training when
8+ was signalled by 81 (filled circles), 82 (open circles),
or 83 (filled squares). Responding to 8- is also presented
(filled triangles). An analysis of variance, performed on
the acquisition data shows that the rate of responding among
the three stimuli did not differ significantly over the last
3 days of training, {S < 1.0.

The right panel of Figure 5 shows the rate of responding
during a white keylight preceded by either 81, 82 or 83

as a function of the interpolated retention interval. The

27
figure shows that 81 (90%) was superior to 82 (50%) and

that 82 was superior to S3 (10%).

A within-subject analysis of variance performed on the
data showed that pre-stimulus and retention interval signifi-
cantly interacted to affect responding during white, F (10,40) =
6.09, p < .001. A test of simple main effects showed that
81 differed from 82 and 83, F (5,40) = 2.94, p < .025 and
F (5,40) = 11.10, p < .001 respectively; and 82 differed from
83, F (5,40) = 4.06, p < .005. Furthermore, a Tukey test with
criterion set at p < .05 showed all pre-stimulus to differ
at 4 sec and 6 sec, and for 81 and 82 to differ from 83
but not each other at 8 sec. The pre-stimuli did not differ
significantly at 2 sec, 10 sec, and 12 sec.

The results clearly demonstrate that the consistency
with which a stimulus predicts 8+ affects its retention.
Furthermore, the poorest predictor of 8+ shows the least
retention, the intermediate predictor intermediate levels
of retention and the best predictor the strongest retention.
The differences in retention implicate memory processes in
the present experiment for the same reasons given in Experi-

ment 2.

GENERAL DISCUSSION

The purpose of the present research was to determine if
the relevance of a stimulus affects its retention. The
results of Experiment 1 showed that food, a stimulus with
direct relevance, was retained for a longer interval than a
stimulus lacking direct relevance. Experiment 2 expanded
this finding to stimuli associated with food. In Experiment 3,
the probability with which a discriminative stimulus predicted
8+ directly affected its retention. Taken as a whole, these
results demonstrate that stimuli differ in their retention
as a function of their relevance to the pigeon. This study
is an initial attempt to determine the relationship between
external stimuli and the animal's memory for these stimuli.
Such an endeavor is potentially useful in furthering our
understanding of both memory and learning processes. For
example, it seems likely that the superior retention found
with food and stimuli associated with food would have some
direct effect on what is learned in an environment when
they are present. In particular, while learning without
reinforcement can occur, superior learning with it may be
more the result of memory factors than motivational factors.

The improved retention observed in these experiments
also raises questions about underlying mechanisms. In general,

the results are consistent with the view that internal

28

29
representations can be achieved by associative stimuli.
Early work by Pavlov (1927) and Konorski (1948) and recent
work by Rescorla and Heth (1975) showed that a substantial
portion of the response to an extinguished CS can be
"reinstated" following separate presentations of the US.
Rescorla and Heth have argued convincingly that modification
of the memory of the U8 account of these findings. Rescorla
and Cunningham (1977) have further demonstrated that rein-
statement can be "erased" following isolated C8 presentation
supporting the view of CS cued retrieval of U8 memories.
An alternative view holds that a direct sensory trace of the
stimulus is formed which then passively decays in a linear
fashion (Roberts & Grant, 1976). This view has difficulty
accounting for the differential retention found in the last
two experiments where all discriminative stimuli were of equal
duration and approximately equal intensity. This is because
the trace decay model assumed that the internal trace is a
direct representation of physical properties of the external
stimulus.

While the present data are consistent with the view
that internal representations are formed, such a concept
does not, in itself, account for the improved retention. An
additional mechanism must be postulated to explain how the
pigeon mediates the retention interval. One popular view is
that stimuli must be actively rehearsed or they will quickly
fade (Grant & Roberts, 1976; Maki, Moe & Bierley, 1977). This

may be viewed as maintenance rehearsal (Atkinson & Shiffrin,

30

1968) which actively maintains information in an accessible,
short term store. The problem with such a model is that
it does not easily account for differences in retention.
If rehearsal is necessary to bridge the retention interval,
all pre-stimuli would have required rehearsal. Therefore,
it is not clear what would lead to differences in retention
between the stimuli. One possibility is that a retrieved
representation of food which is likely to have occurred in
Experiments 1 and 2 facilitated the rehearsal process. This
view, however, would be contrary to the research (Terry &
Wagner, 1975) which suggests that C8 and U8 cued retrievals
result in a suppression in rehearsal. Furthermore, differences
in retention were observed in Experiment 3 in which retrieval
of an internal representation of food was unlikely.

Another hypothesis has been forwarded by Honig (1978)
to explain performance in which a ”working memory" is
necessary. Such a memory is required in paradigms such as
the one used in the experiments presented here, in which
discriminative cues terminate before the animal has the
opportunity to make the correct response. In such cases,
Honig has suggested that the discriminative stimulus
establishes an "instruction" about the criterion response.
This "instruction" is a memory of what is to be done; i.e.,
to peck or not to peck a key. Such an instruction does not
decay over time but is terminated following the opportunity
to execute the criterion response. In the present experiment,
the discriminative stimulus gives the instruction to peck the

key when it is illuminated with the white stimulus. Without

31
the discriminative pre-stimulus the instruction is: don't
peck. One problem with this model is the assumption that
the instruction is either formed at full strength or it
isn't formed. Again, all stimuli were discriminated and
therefore, the pigeon would be expected to form the instruc-
tion to peck for each stimulus.

In the above models, it has been assumed that informa-
tion either passively decays from a sensory store or is
actively maintained in a short term store. An alternative
view is that information is accessible in memory but need '
not be rehearsed. This would be similar to the notion of
”working memory" in humans proposed by Baddeley and Hitch
(1974) and require only some modification in Honig's (1978)
or Olton's (1978) "working memory" models. A trial begins
with retrieval of information about the pre-stimulus, its
associations and the to be executed response. The extent
of information retrieved depends on interference factors
at retrieval and the elaborateness of initial encoding.

The elaborateness of encoding, in part, is a function of

the relevance of the stimulus to the organism. Stimuli

with direct relevance elicit maximum encoding, possibly as

a natural consequence of their importance to the organism.
Encoding of stimuli with associative relevance is also
elaborate and includes the internal representation of a
directly relevant stimulus. Finally, the encoding of stimuli
with discriminative relevance is limited to the "to be
executed" response. It has been demonstrated in human

memory studies that the elaborateness of encoding affects

32
retention of words. Craik and Lockhart (1972) found that
requiring subjects to encode words semantically led to greater
retention than encoding phonemically which, in turn, was
superior to encoding orthographically. A simple explanation
of the present findings is that all stimuli were encoded
for an "instruction" which was sensitive to interference
from alternative behaviors during the retention interval.
In Experiment 3, interference was greatest for 83 (10%)
because the majority of trials during training ended in
non-reinforcement. Behaviors engaged in by the pigeon on
these non-reinforced trials might produce maximum interference
because they involve behaviors directed away from the task
designated by the instruction. Similarly 82 (50%) produced
intermediate interference, while 81 (90%) showed the least
interference. In addition to an encoded "instruction",
the food and C8 pre-stimuli included internal representations
of food. Such memories are less sensitive to interference
from behaviors because they are memories of stimulus events,
not memories of responses. If at the end of a trial, the
pigeon is unable to retrieve the instruction, retrieval of
the internal representation of food can either aid retrieval

of the instructions or re-instruct the pigeon.

LIST OF REFERENCES

LIST OF REFERENCES'

Atkinson, R.C. and Shiffrin, R.M. Human memory: A proposed
system and its control processes. In K.W. Spence and
J.T. Spence (Eds.), The Psychology of Learning and
Motivation: Advances in Research andTTheory, V61. II.
New York: Academic Pfess, 1968, 89-195}

 

Baddeley, A.D. and Hitch, G. Working memory. In G.A. Bower
(Ed.), The Psychology of Learning and Motivation
(Vol. 877’ New York: Academic Press, 1974.

Capaldi, E.J. A sequential hypothesis of instrumental
learning. In K.W. Spence and J. Spence (Eds.),
The Psychology of Learning and Motivation, Vol. I.
New York: Academic Press, 1967.

Capaldi, E.J. Memory and learning: A sequential viewpoint.
In W.K. Honig and P.H.R~ James (Eds.), Animal Memory.
New York: Academic Press, 1971.

Capaldi, E.J., and Cogan, D. Magnitude of reward and differ-
ential stimulus consequences. Psychological Reports,
1963, 13, 85-86.

Cogan, D. and Capaldi, E.J. Relative effects of delayed
reinforcement and partial reinforcement on acquisition
and extinction. Psychological Reports, 1961, 9,
7-13.

Cowles, J.T. and Nissen, N.W. Reward-expectancy in delayed
responses of chimpanzees. Journal of Comparative
Psychology, 1937, 24, 345-358. .

 

Craik, F.I.M. and Lockhart, R.8. Levels of processing:
A framework for memory research. Journal of Verbal
Learning and Verbal Behavior, 1972, 11, 671-684.

Crespi, L.P. Quantitative variations of incentive and
performance in the white rat. American Journal of
Psychology, 1942, 5:, 467-517.

 

Grant, D.8. and Roberts, W.A. Sources of retroactive inhib-
ition in pigeon short-term memory. Journal of Experi-
mental Psychology: Animal Behavioral Processes, 1976,
2, 1-16.

33

34

Honig, W.K. Studies of working memory in the pigeon. In
H. Hulse, H. Fowler, and W.K. Honig (Eds.), Cognitive
Processes in Animal Behavior. Hillsdale: Lawrence
ErIbaum Associates, 1978.

 

 

Konorski, J. Conditioned Reflexes and Neuron Organization.
Cambridge: Cambridge University Press, 1948.

 

Lett, B.T. Delayed reward learning: Disproof of the
traditional theory. Learning and Motivation, 1973,
4, 237-246.

 

Lett, B.T. Long delay learning in the T-maze. Learning and
Motivation, 1975, 6, 80-90.

 

Maki, W.8., Moe, J.C., and Bierley, C.M. Short-term memory
for stimuli, responses, and reinforcers. Journal of
Experimental Psychology: Animal BehavioraI Processes,
1977: 3, 156-I77.

 

 

Olton, D.8. Characteristics of spatial memory. In H. Hulse,
H. Fowler, and W.K. Honig (Eds.), Cognitive Processes
in Animal Behavior. Hillsdale: Lawrence Erlbaum
Associates,71978.

 

 

Pavlov, I.P. Conditioned Reflexes. Oxford: Oxford University
Press, 1927.

 

Rescorla, R.A. Some implications of a cognitive perspective
on Pavlovian conditioning. In H. Hulse, H. Fowler,
and W.K. Honig (Eds.), Cognitive Processes in Animal
Behavior. Hillsdale: Lawrence ErIbaum.Associates,

Rescorla, R.A. and Heth, C.D. Reinstatement of fear to an
extinguished conditioned stimulus. Journal of Experi-
mental Psychology: Animal Behavioral Processes, 1975,
104, 88-96.

Rescorla, R.A. and Cunningham, C.L. The erasure of rein-
stated fear. Animal Learning and Behavior, 1977,
5. 386-394.

Reynolds, G.8. and Catenia, A.C. Temporal discrimination in
pigeons. Science, 1962, 135, 314-315.

Revusky, 8. The role of interference in association over a
delay. In W.K. Honig and P.H.R. James (Eds.),
Animal Memory. New York: Academic Press, 1971.

 

Roberts, W.A. and Grant, D.8. Some studies of short-term
memory in the pigeon with presentation time precisely
controlled. Learning and Motivation, 1974, 5,
393-408.

35

Roberts, W.A. and Grant, D.8. Studies of short-term memory

Spear,

in the pigeon using the delayed matching-to-sample
procedure. In D.L. Medin, W.A. Roberts, and R.T.
Davis (Eds.), Processes of Animal Memory. Hillsdale:
Lawrence Erlbaum Associates, 1976.

 

N.E. Retrieval of memory in animals. Psychological
Review, 1973, 80, 163-164.

 

Sperling, G. A model for visual memory tasks. Human

Factors, 1963, 5, 122-153.

Staddon, J.E.R. Temporal control, attention, and memory.

Terry,

Psychological Review, 1974, 81, 375-391.

W.8. and Wagner, A.R. Short term memory for
"surprising" versus "expected" unconditioned stimuli
in Pavlovian conditioning. Journal of Experimental
Psychology: Animal Behavioral Processes, 1975, 1,
122-153.

Wagner, A.R. Stimulus validity and stimulus selection in

associative learning. In N.J. Mackintosh and W.K.
Honig (Eds.), Fundamental Issues in Associative
Learning. Halifax: Dalhousie University Press, 1969.

Wagner, A.R., Rudy, J.W., and Whitlow, J.W. Rehearsal in

animal conditioning. Journal of Experimental Psychology,
1973, 91, 407-426 (Monograph).

"IIIIIIII'IIIIIIIIII