Citation
Effects of the Introduction of a Memory Aid on Performance in a Modified Delayed Matching to Sample Procedure

Material Information

Title:
Effects of the Introduction of a Memory Aid on Performance in a Modified Delayed Matching to Sample Procedure
Series Title:
Journal of Undergraduate Research
Creator:
Cassidy, Rachel
Hackenberg, Timothy ( Mentor )
Place of Publication:
Gainesville, Fla.
Publisher:
University of Florida
Publication Date:
Language:
English

Subjects

Genre:
serial ( sobekcm )

Notes

Abstract:
The present experiment explored the effects of an observing response, or “memory aid,” in a delayed matching-to-sample procedure with pigeons. Trials began with the onset of a sample stimulus (one of two colors on the center key). After a delay (retention interval), two comparison stimuli were presented on side keys, only one of which matched the sample color. Responses on the matching stimulus were reinforced with food. A third (observing) key was made available when the comparison colors were shown, responses on which reproduced the sample color. This effectively converted the delayed matching-to-sample procedure to a simultaneous matching-to-sample procedure. The response requirement on this observing key increased according to a progressive-ratio schedule that began each session at its minimum of 1 response and increased in fixed increments each time the sample stimulus was reproduced. The retention interval was manipulated systematically across conditions to quantify the tradeoff between the costs of remembering and using a memory aid. Accuracy declined with increases in the retention interval. Observing responses rarely occurred spontaneously, but increased in frequency with additional training on simultaneous matching-to-sample trials. The results are discussed in relation to the training histories required to establish observing in a remembering context.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.

Downloads

This item has the following downloads:


Full Text



Effects of the Introduction of a Memory Aid on Performance in

a Modified Delayed Matching to Sample Procedure


Rachel Cassidy


College of Liberal Arts and Sciences, University of Florida


The present experiment explored the effects of an observing response, or "memory aid," in a delayed matching-to-sample procedure
with pigeons. Trials began with the onset of a sample stimulus (one of two colors on the center key). After a delay (retention
interval), two comparison stimuli were presented on side keys, only one of which matched the sample color. Responses on the
matching stimulus were reinforced with food. A third (observing) key was made available when the comparison colors were shown,
responses on which reproduced the sample color. This effectively converted the delayed matching-to-sample procedure to a
simultaneous matching-to-sample procedure. The response requirement on this observing key increased according to a progressive-
ratio schedule that began each session at its minimum of 1 response and increased in fixed increments each time the sample stimulus
was reproduced. The retention interval was manipulated systematically across conditions to quantify the tradeoff between the costs of
remembering and using a memory aid. Accuracy declined with increases in the retention interval. Observing responses rarely
occurred spontaneously, but increased in frequency with additional training on simultaneous matching-to-sample trials. The results
are discussed in relation to the training histories required to establish observing in a remembering context.


Every act of remembering involves discrete parts:
A to-be-remembered stimulus is presented, and then taken
away during a delay of some kind (Jans & Catania, 1980).
An organism is said to remember the stimulus when its
behavior is shown to be under the control of the stimulus
after the delay. Thus, remembering can be characterized as
a form of delayed stimulus control (Branch 1977).
There are several methods that have been
successfully used to study remembering in animal research,
the most common being the Delayed Matching-to-Sample
(DMTS) procedure (Blough 1959). DMTS is a variation of
the simultaneous matching-to-sample procedure in which a
sample stimulus appears simultaneously with both an
identical comparison stimulus and a comparison stimulus
that differs from the sample. Responses on the comparison
stimulus that matches the sample stimulus are reinforced.
In a typical DMTS procedure, the comparisons are
presented only after a delay (retention interval) during
which neither the sample nor the comparisons are present.
The task of the organism is to choose the matching
comparison.
Many researchers have noted that accuracy in
choosing the correct comparison stimulus decreases as the
retention interval increases (Blough 1959). Increasing the
retention interval also often occasions the appearance of
what has been termed mediating behavior, behavior during
the retention interval that is said to bridge the delay
between the offset of the sample and the onset of the
comparisons. These findings support the intuition that as


the sample recedes farther away in time, the more
cognitively effortful the act of remembering becomes.
Humans, for example, are often presented with information
that must be recalled at a time that is temporally distant
from the present, for example, when coworkers schedule an
appointment for the following week. To enhance recall,
one might use memos, post-it notes, and e-mail reminders;
all are examples of human mediating behavior. When
remembering information for only a short period of time,
however, we often do not resort to these methods because
doing so would actually increase the total effort involved in
the task. A good example of this is looking up a phone
number and then dialing immediately, as opposed to taking
the trouble to find a place to write it down. Clearly, there
are tradeoffs between recall accuracy and the costs of
mediating behavior.
The present research explores the effects of
presenting a pigeon with the opportunity to engage in an
accuracy-enhancing mediating behavior by modifying the
DMTS procedure to include a memory aid. Pigeons were
initially trained on a typical DMTS procedure: the pigeon
was placed in a square chamber with three response keys
mounted on one wall at eye level. The center key was
illuminated either green or white (the sample color) and,
after a delay in which all keys were dark, the two side keys
were both illuminated, one green, the other white.
Following initial training a modification was introduced
such that at the time that the two side keys were
illuminated, the center key was also illuminated red.


University of Florida I Journal of Undergraduate Research I Volume 10, Issue 3 I Spring 2009
1





RACHEL CASSIDY


Responding on this center (observing) key reproduced the
sample color according to a progressive-ratio (PR)
schedule in which the number of responses required to
reproduce the sample increased by five responses
following each completed requirement. The retention
interval was also increased across conditions in order to
make the task more difficult. As noted before, with the
increasing delay, accuracy in choosing the correct
comparison color usually decreases. The introduction of
the center key as a 'memory aid' or 'hint' key serves as an
alternative to remembering. This memory aid, however,
becomes increasingly more costly as the response
requirement heightens. The objective was to better
understand the tradeoffs between remembering and the use
of a memory aid as the costs of each were systematically
varied.


METHOD

Subjects
Four White Cameau pigeons, maintained at
approximately 85% of their free-feeding weights, served as
subjects. The pigeons were housed in a colony room in
individual cages where they had free access to water and
grit. The colony was kept on a light/dark cycle with a light
duration of 16.5 hours.

Apparatus
Sessions were conducted in a standard three-key
operant conditioning chamber for pigeons. The chamber
contained an aluminum intelligence panel with three keys
arranged horizontally. The experimental space measured
30.5 cm x 35 cm x 35 cm. Keys were arranged 23.5 cm
from the floor of the chamber and were spaced 5.7 cm
apart. Each key required a force of 0.12 N to register a
response. Above the center key was a houselight for
general illumination. Below the center key, near the floor,
was a square aperture through which access to mixed grain
could be obtained from a solenoid-operated hopper. All
experimental events were controlled and data collected by
a standard PC using MED-PC IV software.

Procedure
Initial Training. Pigeons were initially trained on a
0-s delayed matching-to sample-procedure. Trials began
with the illumination of a red trial-initiation stimulus on the
center key. One response on the center key changed it to
either green or white. These two colors served as samples,
and occurred with equal probability following a trial-
initiation response. Five responses on the sample key
turned off that key and illuminated the two side keys
(comparisons), one green and one white. The left-right
position of the comparison colors was determined
randomly each trial. A response on the key whose color
matched the sample was reinforced with 3-s access to


mixed grain. An incorrect response was followed by a 10-s
blackout in which all keys and the houselight went dark.
All trials were followed by a 10-s inter-trial interval, during
which the houselight rapidly flashed on and off at a rate of
0.5 s. A correction procedure was also used during initial
training, wherein an incorrect response forced a repeat of
the same trial configuration until the correct comparison
key was pecked. Sessions were conducted daily and were
terminated after either 48 reinforced trials or after 75
minutes had elapsed, whichever occurred first. Subjects
were moved to the experimental phase once they had
completed 5 consecutive sessions with at least 85%
accuracy (correct trials/total trials). Table 1 shows the
sequence and number of sessions per condition.
Experimental Phase. Following initial training, the
retention interval was manipulated across conditions
according to a geometric sequence (1, 2, 4, 8, and 16 s).
Like initial training, 5 responses on the center sample key
turned off that key and initiated the retention interval.
Unlike initial training, the center ('hint') key was
illuminated red simultaneously with the two side-key
comparisons. Pecks on this key reproduced the sample
color associated with that trial type according to a
progressive ratio schedule with a step size of 5 responses
and an initial response requirement of one response. That
is, the ratio requirement increased by 5 responses each time
the sample was reproduced. Correct responses on the
comparison keys continued to be reinforced with food,
irrespective of responses on the center ("hint") key. The
10-s ITI used in initial training continued, but the
correction procedure was discontinued.
Each condition consisted of a minimum of 14
sessions, and continued until accuracy levels were stable.
Stability criteria consisted of taking the mean of the last
nine sessions of a condition and comparing the means of
the first, middle, and last 3 blocks of sessions of those nine.
Means of each three blocks of sessions were not to be more
than 5% different than the overall mean or from each other,
nor could there be evidence of a trend in the means. Table
1 lists the number of sessions per condition for each
subject.

Additional training
Because behavior did not reliably contact with
contingencies in place on the center 'hint' key, further
training procedures were implemented.
Training Phase 1. Training Phase 1 was identical
to the experimental phase described above, with the
addition that, during the first 24 trials within a session, an
FR 1 response requirement replaced the PR schedule on the
center 'hint' key. Completing this FR 1 requirement
changed the center key from red to the sample color
associated with that trial configuration. Prior to the
completion of this requirement, pecks to the comparison
keys received no consequences. The requirement was lifted
after the first 24 trials, and the remaining trials


University of Florida I Journal of Undergraduate Research I Volume 10, Issue 3 I Spring 2009
2





EFFECT OF MEMORY AID

Table 1. Sequence of conditions and number of sessions per condition for each subject.

Condition Subject
009 096 289
Os Retention Interval 35 35 35
Is Retention Interval 30 15 34
2s Retention Interval 23 22 17
4s Retention Interval 19 18 24
8s Retention Interval 14 28 31
16s Retention Interval N/A N/A 16
Training Phase 1 28 19 N/A
8s Retention Interval (2) N/A 16 N/A
Training Phase 2 67 39 64
8s Retention Interval (3) N/A 19 N/A


in the session were identical to the experimental phase.
Because this procedure was insufficient to boost accuracy,
an additional training phase was introduced.
Training Phase 2. Training Phase 2 was identical
to Training Phase 1, with the addition of a correction
procedure in place during the first 24 trials in which an FR
1 response requirement was in place on the center 'hint'
key. Incorrect responses on any of these first 24 trials
forced a repeat of the same trial configuration.

RESULTS

Figure 1 shows overall accuracy (% correct) and accuracy
by trial type as a function of retention interval for each
pigeon. Accuracy decreased as retention interval increased,
with the most dramatic decreases occurring at retention
intervals beyond 4 s. The accuracy for each trial type is
also tracked in order to provide information on the
determinants of accuracy decrements. Trial types 1 and 2
both had a green sample, with a left and right comparison
color, respectively; trial types 3 and 4 had a white sample
with left and right the comparison color, respectively. For
Pigeon 096 in the 8-s retention interval condition, the high
accuracy on trial types 1 and 3 with an overall accuracy of
50% indicates that the pigeon was choosing the left
comparison key regardless of the sample color. Because
the trial types were chosen randomly, there is a 0.5
probability that the correct comparison color would appear
on the left, resulting in the overall accuracy of 50%.
Pigeon 289 showed a similar pattern in choosing the right
key during the 16-s retention-interval condition.
The experimental phase did not consist of forced
exposure to the contingency associated with the 'hint' key,
though during the pre-training experimental conditions


each pigeon did contact the hint key on several occasions.
The initial response requirement was FR 1 and each subject
reproduced the sample a number of times across
conditions, with at least one pigeon (289) having a steady
break point on the center key PR schedule. No pigeon
reliably reproduced the sample more than once per session,
however; that is, they did not complete the next-highest
ratio of 6 after the initial response requirement was met.
This pattern continued despite a decrease in accuracy at
higher retention-interval values. To ensure contact with
the contingency, Training Phase 1 was initiated, which
required pigeons to reproduce the sample on an FR 1
schedule during the first 24 trials of each session before
pecks to the comparison keys could receive consequences.
The pigeons were not reliably more accurate, however, on
those trials during which the sample had been reproduced.
The low accuracy when using the observing key
was perhaps due to the fact that the pigeons had never been
exposed to a simultaneous match-to-sample procedure.
Training Phase 2 was therefore initiated to train
simultaneous matching-to-sample performance. This phase
consisted of the addition of a correction procedure in which
a peck to an incorrect comparison key on any given trial
caused a repeat of that trial type until the correct
comparison key was chosen. The reproduction of the
sample transforms the trial into a simultaneous matching-
to-sample trial (the sample appears simultaneously with the
comparisons). During both training phases, the response
requirement on the center 'hint' key remained at FR 1
during the remaining trials in order to facilitate use of the
'hint' key.
Figure 2 shows the effects of training on accuracy
on only the trials in which the sample had been reproduced
on the center key. Pigeons 009 and 096 were exposed to


University of Florida I Journal of Undergraduate Research I Volume 10, Issue 3 I Spring 2009
3





RACHEL CASSIDY


-1 0 1 2 3 4 5 6 7 8
096












A

A


-1 0 1 2 3 4 5 6 7 8
289
I-A


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14


Retention Interval (s)


Fig.1. Percent Overall accuracy and accuracy by
trial type as a function of retention interval.


15 16 17


-- Total Accuracy
STrial Type 1 Accuracy
A Trial Type 2 Accuracy
a Trial Type 3 Accuracy
A Trial Type 4 Accuracy


University of Florida I Journal of Undergraduate Research I Volume 10, Issue 3 I Spring 2009
4


. . . . . . . . . . . . .


f 009

13a


O~aAL





EFFECT OF MEMORY AID


1001 096


I


60


40


S20



I-J
0
E


I-


100
S80

100
0.


60


40


20


0


Figure 2. Percent correct on trials when the sample has been reproduced. Error bars
represent one standard deviation above and belowthe mean.


both Training Phase 1 and Training Phase 2. For these
two subjects the effect of Training Phase 2 is apparent
in the large increase in accuracy when using the
observing key. In contrast to the other subjects,
Pigeon 289 did not receive Training Phase 1, so the


effects of Training Phase 2 are compared to the 16-s
retention interval condition, which occurred
immediately prior to Training Phase 2. Accuracy
when the sample had been reproduced increased after
Training Phase 2.


University of Florida I Journal of Undergraduate Research I Volume 10, Issue 3 I Spring 2009
5


1001 009


S Training Phase 1
009 -- Training Phase 2

6 I Training Phase 1
-- Training Phase 2


289 l 16 s Retention Interval
l Training Phase 2


289


-+





RACHEL CASSIDY


1W 1 096


I


80


60


40


*o

E


I-


61M
0

V 80
I-
d)
60


40


20


0


Figure 2. Percent correct on trials when the sample has been reproduced. Error bars
represent one standard deviation above and belowthe mean


DISCUSSION

The effect of the introduction of a memory aid in
this modification of a DMTS procedure is difficult to
assess given that the observing key was not consistently
pecked prior to training. Although overall accuracy
decreased as a function of retention interval length, the
pigeons did not spontaneously reproduce the sample.
Training phases initiated later in the experiment arranged
the first half of each session such that pigeons were forced
to reproduce the sample on an FR 1 schedule before pecks
to either of the comparisons could receive consequences.
The lack of improved accuracy on those trials during which
the sample had been reproduced provides evidence that the
memory aid key was not actually functioning as a memory


aid in human terms; that is, the reproduction of the same
sample color evidently failed to 'remind' the subjects of
the sample.
The performance seen under this modification of
the DMTS procedure raises important issues regarding
stimulus control by the sample in DMTS procedures, as
well as what subjects are remembering when accuracy
levels are high. In a functional analysis of DMTS findings,
Wixted (1989) hypothesized that the efficacy of the sample
as a discriminative stimulus is related to its ability to
function as a conditioned reinforcer. In the context of
DMTS procedures, Wixted modified the delay-reduction
hypothesis-which indicates that conditioned reinforcers
gain efficacy through their correlation with a reduction in
the delay to the availability of reinforcement-to include


University of Florida I Journal of Undergraduate Research I Volume 10, Issue 3 I Spring 2009
6


100 009


J Training Phase 1
009 Training Phase 2

0 Training Phase 1
096 Training Phase 2


289 16s Rtention Interval
STraiinng Phase 2


289 T


I 1 1 I


I I 1 I


rI?





EFFECT OF MEMORY AID


the effects of the retention interval. This model predicts
that when the retention interval is large or the delay
reduction associated with the sample is small, the sample
stimulus will exert little control as a discriminative
stimulus. He postulated that in such a context the
comparison stimuli will function as discriminative stimuli
independently of the sample. This is an important
consideration for interpreting the results of the present
study.
Consistent with the stipulations of Wixted's model
that would predict the independence of the comparison
stimuli as discriminative stimuli, the onset of the sample in
the present experiment does not correlate with a
meaningful reduction in the delay until reinforcement is
available. Additionally, when the retention interval was
large, biases developed for two pigeons (096 and 289)
which began to respond at chance levels. When taking the
comparison stimuli as independent of the sample, however,
then each key has an equal probability of reinforcement, so
the subject still earns an average of 50% of the available
reinforcement through chance responding. More
importantly, the pigeons did not reliably use the observing
(memory aid) key without explicit training, though
effective use could have ensured 100% of the available
reinforcement.
When the pigeons were exposed to Training Phase
1 in which they were forced to reproduce the sample before
they could peck either of the comparison keys, their
accuracy on the trials in which the sample was reproduced
was low (Figure 2). This indicates that the memory aid key
was not actually functioning as a memory aid in human
terms. The explanation for this discrepancy may lie in the
way that DMTS trials are learned. When the retention
interval is short, and the sample presumably retains its
discriminative function, it is unlikely that the pigeons were
matching the color green with the color green; in other
words, both the sample and the configuration of the
comparisons as they appear after the sample are learned as
discrete events. The differential accuracies by trial
configuration (Figure 1) support this conclusion. When
there is a short retention interval, one would predict that
the pigeons would be unlikely to make use of a memory
aid, which indeed they did not. As the retention interval
increased, however, and remembering the trial
configuration presumably became more difficult, then the
introduction of a third key light simultaneously with the
comparisons might be expected to disrupt rather than to
enhance performance.
If the diminished stimulus control by the sample
during longer retention-interval conditions, supported by
chance responding, is evidence that the pigeons no longer
remembered the sample, then it is unlikely that a memory
aid that reproduces the sample color later on in the trial
would simply 'remind' the pigeons of the sample in the
absence of explicit pairing of the two stimuli. The
subsequent training phases that improved accuracy when


using the memory aid suggests that, with proper pairing,
pigeons can utilize the memory aid key in a fashion
analogous to the manner in which humans use memory
aids.
In conclusion, the results of the present
experiment provide evidence that the mere opportunity to
reproduce the sample color did not function as a memory
aid without explicit training. Future experiments using this
procedure should begin with simultaneous match-to-
sample training to facilitate performance during trials when
the sample has been reproduced and appears
simultaneously with the comparisons. Additionally, future
experimentation using this procedure should include
explicit training of the observing response, preferably with
a correction procedure in place to ensure sufficient pairings
of the original and reproduced sample colors. Such
experimentation is currently underway.

REFERENCES

Berryman, R, Cumming, W.W., & Nevin, J.A. (1963). Acquisition of delayed
matching in the pigeon. Journal of the Experimental Analysis of
Behavior, 6, 101-107.

Blough, D.S. (1959). Delayed Matching in the Pigeon. Journal of the
Experimental Analysts ofBehavior, 2, 151-160.
Branch, M.N. (1977). On the role of 'memory' in the analysis of behavior.
Journal of the Experimental Analysts ofBehavior, 28, 171-179.
Cumming, W.W. & Berryman, R. (1965). The complex discriminated operant:
Studies of matching-to-sample and related problems. In D.D.
Mostofsty (Ed.) Stimulus Generalization (pp. 284-330). Stanford, CA:
Stanford University Press.
Dinsmoor, J.A. (1985). The role of observing and attention in establishing
stimulus control. Journal of the Experimental Analysis ofBehavior,43,
365-381.
Jans, J.E., and Catania A.C. (1980). Short-term remembering of discriminative
stimuli in pigeons. Journal of the Experimental Analysis of Behavior,
34, 17-183.

Mackay, H.A. (1991). Conditional stimulus control. In Iverson, I.H. & Lattal,
K.A. (Eds.) Experimental Analysts ofBehavior, Part 1 (pp. 301-350).
New York: Elsevier.

Nevin, J.A., Davison, M., & Shahan, T.A. (2005 ). A theory of attending and
reinforcement in conditional discrimination. Journal of the
Experimental Analysts ofBehavior, 84, 281-303.
Reynolds, G.S. Attention in the Pigeon. Journal of the Experimental Analysis of
Behavior, 4, 203-208.
Shimp, C.P. & Moffitt, M. (1974). Short-term memory in the pigeon: Stimulus-
response associations. Journal of the Experimental Analysis of
Behavior, 22, 507-512.
Shimp, C.P.(1976). Short-term memory in the pigeon: Relative recency. Journal
of the Experimental Analysts ofBehavior, 25, 55-61.
Stafford, D. & Branch, M.N. (1998). Effects of step size and break-point criterion
on progressive-ratio performance. Journal of the Experimental
Analysis ofBehavior, 70, 123-138.
White, K. G. (1985). Characteristics of forgetting functions in delayed matching
to sample. Journal of the Experimental Analysis of Behavior, 44, 15
34
Wixted, J.T., 1989. Nonhuman short-term memory: a quantitative reanalysis of
selected findings. Journal of the Experimental Analysis of Behavor,
52, 409-426.


University of Florida I Journal of Undergraduate Research I Volume 10, Issue 3 I Spring 2009
7