Journal of ExDerimental Psychology: Copyright 1981 by the American Psychological Association, Inc.

Animal Behavior Processes 0097-7403/81 /0703-0217S00.75

1981, Vol. 7, No. 3, 217-227

Intertrial Interference in Rat Short-Term Memory

Douglas S. Grant
University of Alberta, Edmonton, Alberta, Canada

The delayed alternation behavior of rats was investigated in four experiments

with a T-maze. Each trial involved an initial forcing of the subject to one arm
of the T, followed either 0 or 40 sec later by a free-choice run on which a turn
•in either direction could be made. In order to obtain a reinforcer on the free-
choice run, the rat was required to turn in the direction opposite that of the
forcing. During experimental testing, delayed alternation target trials were pre-
ceded by an initial delayed alternation trial which was either the same as or
opposite to the target trial. Relative to a control in which an initial trial was not
presented, accuracy on the free-choice run of target trials was reduced by either
type of initial trial (interference effect). Experiments 1 and 2 demonstrated that
(a) interference was obtained when the target-trial retention interval was 40 sec
but not when it was 0 sec and (b) interference was reduced by repetition of the
initial trial and by increased temporal separation between the initial and the
target trials. Experiment 3 demonstrated that an initial trial that consisted en-
tirely of forced runs produced an interference effect at both the 0- and 40-sec
retention intervals. The findings are discussed in relation to previous studies of
interference effects in rats and pigeons.

Several recent investigations of rat short- permits the study of the effects of prior in-
term memory have employed a delayed al- formation on the retention of subsequent in-
ternation procedure (Feldman & Gordon formation. In this preparation, prior infor-
1979; Grant, 1980; Roberts, 1974). In the mation is introduced by initially forcing the
procedure employed, the rat is initially forced animal to the side opposite that of the target
to one side of a T-maze (called the target forcing (called the interfering forcing). Fol-
forcing) and, following a retention interval lowing administration of the target forcing
of several seconds, is released for a free- and passage of the retention interval, the rat
choice run. The animal is rewarded on the is released on a free-choice run. A reinforcer
free-choice run only if it turns in the direc- is present only in the goal box of the arm
tion opposite that of the target forcing. Suc- opposite that to which the animal was forced
cessful delayed alternation performance thus during the target forcing, and a turn in this
requires the rat to remember the direction direction is defined as a correct response. To
of the target forcing and to use this memory obtain a reinforcer, the rat must thus base
to control choice responding. the choice response on the memory of the
Roberts (Note 1) and Gordon, Brennan, more recent, target forcing.
and Schlesinger (1976) developed a variant Several studies have shown that such in-
of the delayed alternation procedure which terfering forcings represent a significant
source of proactive interference, leading to
a reduced level of free-choice accuracy (Gor-
don et al., 1976; Gordon & Feldman, 1978;
The research was supported by Grant A0443 from
the Natural Sciences and Engineering Research Council Grant, 1980; Roberts, Note 1). These studies
of Canada. The technical assistance of Linda Scharr is have shown further that the magnitude of
gratefully acknowledged. The data were reported at the proactive interference is independent of
meeting of the Canadian Psychological Association, length of retention interval; that is, the pres-
Calgary, Alberta, June 1980.
Requests for reprints should be sent to Douglas S. ence of interfering information leads to an
Grant, Department of Psychology, University of Al- equivalent decrease in percentage of correct
berta, Edmonton, Alberta T6G 2E9, Canada. free-choice responses on immediate (i.e., 0-

217
218 DOUGLAS S. GRANT
sec retention interval) and delayed (e.g., 40-
sec retention interval) tests.
The present experiments investigated fur-
ther the effect of prior information on the
retention of subsequent information in rat
short-term memory. A delayed alternation
procedure was employed in which an entire
delayed alternation trial (i.e., forcing plus
free-choice run), rather than only an initial
forcing, was administered as the interfering
event. The present preparation may be re-
ferred to as an intertrial interference pro-
cedure, whereas that developed by Roberts
(Note 1) and Gordon et al. (1976) may be
referred to as an /nfnztrial interference pro-
cedure.
Experiment 1
In the first experiment, retention on target
trials was tested at intervals of 0 and 40 sec
in both the presence (interference trials) and
absence (control trials) of an immediately
prior delayed alternation trial. On interfer-
ence-same trials, the interfering and target
trials were identical, whereas on interfer-
ence-different trials, the interfering trial was
opposite to the target trial (see Table 1). In
each of the interference conditions, the ini-
tial trial was administered either one or four
times prior to the target trial.
Method
Subjects. Twelve naive male Sprague-Dawley rats,
approximately 60 days old upon arrival in the labora-
tory, were used. Subjects were reduced to and main-
tained at 80% of their ad lib weight.
Apparatus. A T-maze, measuring 10 cm wide with
sides 13 cm high, was used. The stem portion was 43
cm long, and each arm was 56 cm long. Clear Plexiglas
guillotine doors were mounted at the entrance to each
arm and were used to control the direction of the turning
response on forced runs. A 15-cm-long goal box was
attached at a right angle, extending back toward the
start box, at the end of each arm. The T-maze was
located in a 4.5 X 4 m room.
Procedure. During alternation training, each animal
received 10 spaced trials per session, 5 left-correct and
S right-correct, in random order. Each trial involved a
reinforced (four 45-mg food pellets, the reinforcer used
throughout) forced run to either the right or the left.
After the reward had been consumed, the rat was re-
turned to the start box and immediately released (0-sec
retention interval) on the free choice. On the free-choice
run, neither arm was blocked, and reward was present
only in the goal box on the side opposite that entered
on the forced run. If the rewarded side was entered, a
correct response was recorded, and the animal was re-
turned to the holding cage following consumption of the
reinforcer. If the nonrewarded side was entered, an in-
correct response was recorded, and the rat was returned
to the start box and allowed an opportunity to enter the
rewarded side. Correction runs were continued until the
correct arm was chosen. Alternation training continued
for 10 days, at which point all animals alternated with
a high degree of accuracy (90% or better).
Following alternation training, the correction proce-
dure was discontinued, and testing on the first experi-
ment began. Animals were tested at retention intervals
of 0 and 40 sec in each of five conditions in a within-
subjects design (see Table 1). In the Control condition,
animals received a delayed alternation target trial in the
absence of an immediately preceding trial. In the In-
terference conditions, the target trial was preceded im-
mediately by the administration of an initial trial. The
initial trial was either identical to the target trial (In-
terference-Same, IS) or opposite to the target trial (In-
terference-Different, ID). In each of the Interference
conditions, the initial trial was given only a single ad-
ministration (IS1 and ID1) or was repeated four times
prior to the target trial (IS4 and ID4). The retention
interval on all initial trials was 0 sec, and the interval
between successive repetitions of the initial trial in the
IS4 and ID4 conditions was 0 sec.
Combining the factors of conditions (Control, IS1,
IS4, ID1, and ID4), retention interval (0 and 40 sec)
and correct position on the free-choice run of the target
trial (left and right) yields 20 unique trial configura-
tions. Animals were tested on five trial configurations
each session, one from each of the five conditions. The
daily sequence in which each condition was tested was
determined randomly for each subject. Once every four
sessions, animals were tested on each of the 20 trial
configurations; over a block of eight sessions, they were
tested on each configuration twice. Animals were run
in rotation in squads of six, and the interval between
successive trial configurations for any particular rat was
approximately 7, plus or minus 2, min. Animals were
tested for four blocks of eight sessions, which produced
a total of 192 observations at each retention interval in
each of the five conditions.
Table 1
An Example Trial Configuration in Each of
the Conditions Employed in Experiment 1
Interference- Interference-
Runs Control Same Different
Initial
Forced none left right
Free none right? left?
Target
Forced left left left
Free right? right? right?
Note. On the other half of the occasions, the target trial
consisted of a forced run to the right and a free-choice
run with left correct. On such trials, the direction of the
forced and free-choice runs on the initial trial was op-
posite those shown.
 INTERFERENCE IN RAT SHORT-TERM MEMORY 219

Results ent) X Repetitions (1 and 4) X Retention

Interval X Blocks ANOVA was performed on
Mean percentage of correct free-choice the four Interference conditions. Significant
runs on the target trial as a function of con- main effects of type, F(l,l 1) = 11.85, p <
ditions and retention interval is shown in .01, and repetitions, F(l, 11)= 15.07, p<
Figure 1. In the left panel, accuracy on the .01, suggest that the IS conditions produced
target trial in IS1 and IS4 is shown in re- greater interference than the ID conditions
lation to Control accuracy, and in the right and that a single repetition of the initial trial
panel, accuracy on the target trial in ID1 produced greater interference than four rep-
and ID4 is shown in relation to Control ac- etitions. The significant Repetitions X Re-
curacy. The Control curve is identical in tention Interval interaction, F( 1,11) = 24.33,
both panels and was included in each to fa- p < .001, suggests that rate of forgetting was
cilitate comparison. faster given a single repetition of the initial
A Conditions X Retention Interval X trial than given four repetitions. The failure
Blocks analysis of variance (ANOVA) re- of the Type X Retention Interval interaction
vealed significant main effects of conditions, to approach significance (F< 1) suggests
/r(4, 44) == 8.98, p < .001, and retention in- that rate of forgetting on the target trial was
terval, F(l,11)= 105.37, p<.001, and a independent of whether the initial trial was
significant interaction between the two, F(4, the same as or different from the target trial.
44) = 8.02, p < .001. The interaction reflects These statements were supported further
the fact that rate of forgetting on the target by the results of two additional analyses.
trial was affected markedly by conditions. First, each of the four Interference condi-
Relative to the Control, IS1 and ID1 ap- tions was compared individually with the
peared to demonstrate more rapid forget- Control. Both of the analyses involving a
ting, whereas IS4 and ID4 appeared to dem- single repetition of the initial trial (Con-
onstrate somewhat less rapid forgetting. trol X IS1 and Control X ID1) revealed a
In order to pursue these effects further, significant Conditions X Retention Interval
a Type of Initial Trial (Same and Differ- interaction, Fs(l, 11) =15.10 and 26.10,

100 Control + Interference • Same

100
Control -I- Interference • Different

90

80

70

60 60
Control Control
lnterrerence-1 Rep lnterference-1 Rep
Interference--) Reps lnterference-4 Reps

50.
40 0 40
Retention Interval (sec)
Figure 1. Mean percentage of correct free-choice runs on the target trial as a function of conditions and
retention interval in Experiment 1.
220 DOUGLAS S. GRANT

respectively, both ps<.01. On the other tablished a separate memory, one of the two
hand, this interaction failed to reach signif- memories would necessarily be in conflict
icance in either analysis involving four rep- with the memory of the forcing on the target
etitions of the initial trial (Control X IS4 trial regardless of whether the initial trial
and Control XID4). Second, two-tailed was designated as "same" or "different."
Dunnett's tests were performed at each re- Similarly, one of the two memories would
tention interval to compare Interference and necessarily be congruent with the memory
Control performance. At the 0-sec interval, of the target-trial forcing. If it is assumed
none of the Interference conditions differed further that the interference resulting from
significantly from the Control, whereas at a conflicting memory is greater than the fa-
the 40-sec interval both IS1 and ID1 were cilitation resulting from a congruent mem-
significantly less accurate than the Control, ory, resolution of the fact that both IS and
ts (5, 44) = 4.22 and 3.48, respectively, both ID resulted in interference may be accom-
ps < .01. plished by viewing the initial trial as com-
posed of two at least partially separable ele-
ments.
Discussion
Consistent with this interpretation is the
The results obtained in the present exper- finding that IS produced somewhat greater
iment employing the intertrial interference interference than ID. In the IS condition it
procedure differ from those obtained in stud- is the more recent free-choice run that con-
ies employing the intratrial interference pro- flicts with the target forcing, whereas in the
cedure in two major respects. First, the in- ID condition it is the more temporally re-
tratrial procedure results in robust and mote forced run that conflicts with the target
comparable levels of interference on both forcing. To the extent that less recent mem-
immediate and delayed tests of retention ories compete less successfully with current
(Gordon et al., 1976; Gordon & Feldman, memories, greater interference in the IS
1978; Grant, 1980; Roberts, Note 1). In con- than in the ID condition would be antici-
trast, the intertrial procedure results in in- pated.
terference only on a delayed test of retention. Experiment 2
A second point of contrast between intratrial The purpose of the second experiment was
and intertrial studies of proactive interfer- two-fold. First, the effect of number of rep-
ence in the rat concerns the effect of repe- etitions of the initial trial on proactive in-
tition of interfering information. Both Gor- terference in the intertrial procedure was
don et al. and Roberts (Note 1) found that explored further by presenting the initial
increasing the number of repetitions of the trial once, three times, or six times prior to
interfering information increased proactive the target trial. On the basis of the outcome
interference on both immediate and delayed of the first experiment, it was predicted that
tests. On the other hand, increased repetition proactive interference would be less appar-
of the initial trial led to a marked reduction ent as repetition of the initial trial increased.
in proactive interference in the present ex- Second, the effect of the degree of temporal
periment. spacing between the final repetition of the
One anomalous aspect of the present data initial trial was assessed. The three levels of
concerns the finding that both IS and ID spacing employed were 0, 30, and 60 sec.
conditions produced interference on the de- Studies of proactive interference in the in-
layed retention test. This anomaly may be tratrial procedure have demonstrated that
more apparent than real, however, when the interference decreases as the degree of spac-
nature of the initial trial is considered. Initial ing between interfering and target infor-
trials consisted of two components, the forced mation increases (Gordon & Feldman, 1978;
run and the free-choice run. Since animals Roberts, Note 1).
alternated with a high probability at the 0-
sec retention interval (see Figure 1), most Method
initial trials involved a run to each arm of Procedure. The subjects and apparatus of the first
the maze. To the extent that each run es- experiment were again used. Approximately 1 wk after
 INTERFERENCE IN RAT SHORT-TERM MEMORY
completion of the previous experiment, animals were
given four sessions of delayed alternation training. Per-
formance was consistently high across sessions, ranging
from 92% to 97% correct.
Testing on Experiment 2 began on the day following
the final session of alternation training. Animals were
divided into two groups of six rats each. Animals in
Group Same-Different experienced Control, IS1, IS3,
and IS6 conditions, whereas animals in Group Different-
Same experienced Control, ID1, ID3, and ID6 condi-
tions. After 24 sessions of testing, the groups were
switched such that Same-Different animals were tested
on Control, ID1, ID3, and ID6, and Different-Same
animals were tested on Control, IS1, IS3, and IS6. Test-
ing continued for an additional 24 sessions following the
switch.
In each of the Interference conditions, the interval
separating the end of the final initial trial repetition and
the onset of the forcing on the target trial (ITI) was 0,
30, or 60 sec. In each block of 24 sessions, each rat was
tested on the 2 control trial configurations and 18 in-
terference trial configurations (3 Repetitions X 3
ITIs X 2 Retention Intervals) appropriate to its group
and phase, receiving either 6 or 7 trial configurations
per session. The daily sequence in which trial configu-
rations were presented was determined randomly for
each subject. Subjects were run in rotation in squads
of six, and the interval between successive trial config-
urations for any particular rat was approximately 9, plus
or minus 3, min. All other aspects of the procedure were
the same as in the first experiment.
Results
Mean percentage of correct free-choice
runs on the target trial as a function of treat-
ments is shown in Figure 2. The left panels
present Control and IS performance, and the
rights panels present Control and ID per-
formance. Control accuracy was assessed
separately, depending upon whether the an-
imals experienced IS or ID conditions in that
block. The Control curve is plotted at each
of the three ITIs to facilitate comparison
among conditions. Since the main effect of
the order in which animals received IS and
ID testing was not significant and the order
factor did not interact with any other factor
or combination of factors, the data were col-
lapsed across order.
A Type of Initial Trial X ITI X Repe-
titions X Retention Interval ANOVA per-
formed on the Interference conditions re-
vealed greater interference in IS than in ID,
F(l, 11) = 6.17, p < .05, decreasing inter-
ference as ITI increased, F(2,22) = 12.27,
p < .001, decreasing interference as repeti-
tions increased, F(2,22) = 4.71, p < .05, and
forgetting across the retention interval, F( 1,
221
11) = 169.39, p< .001. The ITI X Retention
Interval interaction, F(2, 22) = 14.93, p <
.001, suggests that manipulation of ITI had
little effect on accuracy on the immediate
test and a marked effect on accuracy on the
delayed test. The Repetitions X Retention
Interval interaction, F(2, 22) = 3.89, p <
.05, similarly suggests that repetitions influ-
enced performance primarily on the delayed
test. Finally, the ITI X Repetitions X Re-
tention Interval interaction, F(4, 44) = 2.94,
p < .05, suggests that all three factors jointly
determine the amount of interference ob-
tained.
Two-tailed Dunnett's tests were per-
formed to compare the three Interference
conditions with the Control at each ITI and
retention interval in both IS and ID condi-
tions. At the 40-sec retention interval and
a 0-sec ITI, IS1, IS3, and ID1 were each
significantly less accurate than the Control,
ts (4, 33) = 3.57, 2.14, and 4.27, respec-
tively, all ps < .05; no other comparisons
were significant.
Discussion
The present findings both replicate and
extend those of the first experiment. As in
the previous experiment, accuracy of free-
choice responding was reduced at a 40-sec
retention interval when the target trial was
immediately preceded by a single repetition
of an initial trial which was either the same
as or opposite to the target trial. Although
different values of number of repetitions of
the initial trial were employed in the prior
and present experiment (1 and 4 vs. 1,3, and
6), repetition markedly attenuated proactive
interference in both experiments.
Finally, manipulation of the interval sep-
arating the terminal initial trial repetition
and onset of the target trial produced sub-
stantial effects. Specifically, no evidence of
proactive interference was obtained when
the ITI was either 30 or 60 sec. This finding
is consistent with those of studies that have
employed the intratrial procedure to study
proactive interference in the rat (Gordon
& Feldman, 1978; Roberts, Note 1).
Experiment 3
As noted earlier, prior work on proactive
interference in the delayed alternation task
222 DOUGLAS S. GRANT

Control + Interference • Same Control + Interference - Different

100 100 ._
ITI = 0 sec ITI = 0 sec
^tC^>^
90 90

80 80

70 70

60 60

SO
• i |
GC " 40 0 40
®
o
100 100
O ITI = 30 sec ITI — 30 sec
0) 90 90
2
80

«"
? 70 70

2»
O
60

S> so
2 SO

40 40
0)
CL
100 100
ITI = 60 sec ITI = 60 sec
90 90

80 80

70 70
•_—• Control • •Control
60 o——o lnterference-1 Rep 60 ) lnterference-1 Rep
* a Interference 3 Reps v Interference-3 Reps
50 o——o lnterference-6 Reps 50 3 lnterference-6 Reps
i i I
> 40 40

Retention Interval (sec)

Figure 2. Mean percentage of correct free-choice runs on the target trial as a function of treatments
in Experiment 2.

employed an intratrial procedure and ob-
tained robust interference on both immedi-
ate and delayed tests of retention (Gordon
et al., 1976; Gordon & Feldman, 1978;
Grant, 1980; Roberts, Note 1). In contrast,
the present work, employing an intertrial
procedure, has demonstrated robust inter-
ference only on a delayed test of retention.
Two essential differences between these pro-
cedures may be identified, either or both of
which may be responsible for the contrasting
results. First, the initial runs are all forced
in the intratrial procedure, whereas in the
intertrial procedure half are forced and half
are free. Second, the initial runs are all to
the same arm of the maze in the intratrial
procedure, whereas in the intertrial proce-
dure they are divided equally between the
two arms, provided the animal alternates on
the free-choice run.
The primary purpose of the final experi-
ment was to determine which of these pro-
 INTERFERENCE IN RAT SHORT-TERM MEMORY
cedural differences is responsible for pro-
ducing different levels of interference on an
immediate test. In order to accomplish this,
animals were tested in an initial phase on
three conditions; Control, IS1, and ID1 at
each of four retention intervals (0, 20, 40,
and 50 sec). The initial phase represented
a partial replication of the first two experi-
ments, with a different sample of subjects
and more extensive sampling of the length-
of-retention-interval variable. In addition,
this phase provided a baseline on which to
assess the influence of treatments employed
in the second phase.
In the second phase, which immediately
followed completion of the first phase, the
animals were tested on four conditions; Con-
trol, IS!', ID1', and 2PF (seeTable 2). Con-
trol trials were identical to those employed
in the previous experiments. Conditions 1ST
and IDT corresponded to IS1 and ID1, re-
spectively, of the first phase with one notable
exception, the second run was also forced in
1ST and IDT, that is, following the initial
forcing, animals were forced to the opposite
arm rather than receiving a free-choice run.
Finally, the 2 PF condition corresponded to
the intratrial procedure, and, therefore, an-
imals received two initial forcings to the
same arm of the maze prior to the target
forcing. In each of the four conditions of the
second phase, animals were tested at reten-
tion intervals of 0 and 40 sec.
To the extent that robust interference on
an immediate test is critically dependent
upon the initial runs occurring in only one
direction, Condition 2PF should produce
substantial interference on both immediate
and delayed tests, while Conditions 1ST and
IDT should produce interference only on a
delayed test. In addition, this hypothesis pre-
dicts that a between-phase comparison be-
tween 1ST and IS1 conditions and between
IDT and ID1 conditions should fail to reveal
differences in retention on an immediate
test,
On the other hand, to the extent that ro-
bust interference on an immediate test is
critically dependent upon all initial runs
being forced, different interference effects
across the retention interval in Conditions
1ST, ID1', and 2PF would not necessarily
be anticipated. This hypothesis also predicts
223
that retention on an immediate test should
be less accurate in 1ST than in IS1 and less
accurate in ID1' than in ID1.
Method
Subjects, Twelve naive male Sprague-Dawley rats,
approximately 60 days old upon arrival in the labora-
tory, were used. Subjects were reduced to and main-
tained at 80% of their ad lib weight.
Apparatus. The apparatus was the same as in Ex-
periment 1.
Procedure. Animals were given eight sessions of al-
ternation training initially, by the procedure used in the
first experiment. Animals were then given six additional
sessions of alternation training with retention intervals
of 0, 20, 40, and 60 sec. Each retention interval was
tested twice each session. The percentages of correct
free-choice runs over the six sessions were 96, 90, 85,
and 75 at the 0-, 20-, 40-, and 60-sec retention intervals,
respectively.
Testing on Phase 1 began immediately following the
final session of training. Animals were tested on three
conditions, Control, IS1, and ID1. In each condition,
four retention intervals were tested, 0, 20, 40, and 60
sec. Each animal received each of the 12 combinations
of condition and retention interval once per session. The
daily sequence in which each combination of condition
and retention interval was tested was determined ran-
domly for each subject. Animals were run in rotation
in squads of six, and the interval between successive trial
configurations for any particular rat was approximately
5, plus or minus 1, min. The first phase was run for 18
sessions, which yielded 216 observations at each reten-
tion interval in each condition.
Immediately following completion of the first phase,
animals began testing on Phase 2. The four conditions
employed are illustrated in Table 2. Control trials were
Table 2
An Example Trial Configuration in Each of
the Four Conditions Employed in the Second
Phase of Experiment 3
Runs Control isr ior 2PF
Initial trial
Forced none left right right
Forced none right left right
Target trial
Forced left left left left
Free right? right? right? right?
Note. On the other half of the occasions, the target trial
consisted of a forced run to the right and a free-choice
run with left correct. On such trials, the direction of the
initial forcings was opposite those shown. 1ST = Inter-
ference-Same, one repetition; IDT = Interference-Dif-
ferent, one repetition. The primes indicate that both
initial runs were forced. 2PF = two prior forcings, both
to the side opposite that of the target forcing.
224 DOUGLAS S. GRANT
identical to those of the first phase. Conditions 1ST and
IDT corresponded to IS1 and ID1, respectively, except
that the animals were forced to alternate on the second
initial run rather than receiving a free choice. Finally,
in the 2PF condition, the intratrial procedure was fol-
lowed in that animals received two initial forcings to the
arm of the maze opposite that of the target forcing. In
each of the four conditions, half of the target trials in-
volved a 0-sec retention interval and half a 40-sec in-
terval.
Animals were tested on 12 trial configurations per
session. The daily sequence in which trial configurations
were tested was determined randomly for each subject.
Animals were run in rotation in squads of six, and the
interval between successive trial configurations for any
particular rat was approximately 5, plus or minus 1,
min. The second phase was run for 18 sessions, which
yielded 324 observations at each retention interval in
each condition.
Results
Mean percentages of correct free-choice
runs on the target trial as a function of con-
ditions and retention interval during Phase
1 are shown in Figure 3. All three conditions
demonstrated forgetting across the retention
interval, with rate of forgetting more marked
in the Interference conditions. A Condi-
tions X Retention Interval ANOVA revealed
significant main effects of conditions, F(2,
22) = 23.66, p < .001, and retention inter-
val, F(3, 33) = 63.59, p < .001. The signif-
icant interaction, F(6, 66) = 3.49, p < .01,
suggests that the more rapid rate of forget-
ting in the Interference conditions was re-
liable. In order to further ensure the reli-
ability of this effect, separate Condi-
tions X Retention Interval ANOVAS were
performed to compare the Control with IS1
and the Control with ID1. In both cases, the
Conditions X Retention Interval interaction
was significant, Fs(3, 33) = 3.03 and 4.92,
respectively, both ps < .05.
Two-tailed Dunnett's tests were per-
formed to compare accuracy in each of the
Interference conditions with that in the Con-
trol condition at each retention interval. At
the 0-sec interval, neither condition differed
from the Control; at the 20-sec interval, only
ID1 differed reliably from the Control, t(3,
22) = 3.05, p < .01. At both the 40- and 60-
sec intervals, both IS1 and ID1 differed re-
liably from the Control, fs(3, 22) = 4.89 and
4.01 at 40 sec, respectively, and 4.23 and
5.37 at 60 sec, respectively, all ps < .05.
lOOr
90
3
CC
BO
70
60-
Control
Interference - Same
Interference • Different
50 T
20 40 60
Retention Interval (sec)
Figure 3. Mean percentage of correct free-choice runs
on the target trial as a function of conditions and re-
tention interval in the first phase of Experiment 3.
Mean percentage of correct free-choice
runs on the target trial as a function of con-
ditions and retention interval during Phase
2 is shown in Figure 4. Performance was
influenced markedly by conditions, F(3,
33) = 36.28, p < .001, and retention inter-
val, F(l, 11) = 46.21, p < .001. All three
Interference conditions demonstrated lower
accuracy than the Control on the immediate
test, Dunnett's test, <s(4, 33) = 2.84, 4.17,
and 8.78 in ID1', IS1', and 2PF, respectively,
allps < .05, two-tailed. Although rate of for-
getting was somewhat slower in the Inter-
ference conditions than in the Control, this
effect did not reach significance, F(3, 33) =
1.74. Newman-Keuls tests were performed
on the data, collapsed across retention in-
terval, and revealed that each condition dif-
fered significantly (ps < .05 in all cases)
from each of the remaining three in the fol-
lowing arrangement, Control > I D l ' >
IS1' > 2PF.
Comparison of Figures 3 and 4 reveals
that free-choice performance on an imme-
diate test was less accurate when both initial
runs were forced (1ST and ID1') than when
one was forced and the other free (IS1 and
ID1). Two-tailed Student's t tests revealed
both of these effects to be significant; 1ST
was less accurate than IS1, t ( l l ) = 2.26,
 INTERFERENCE IN RAT SHORT-TERM MEMORY
100r
X
X
40
Retention Interval (sec)
Figure 4. Mean percentage of correct free-choice runs
on the target trial as a function of a conditions and
retention interval in the second phase of Experiment 3.
(1ST = Interference-Same, one repetition; IDT = In-
terference-Different, one repetition. The primes indicate
that both initial runs were forced. 2PF = two prior forc-
ings, both to the side opposite that of the target forcing.)
p < .05, and IDT was less accurate than
ID1, t ( l l ) = 2.53, p< .05.
Discussion
The results of the first phase replicated
the primary finding of the first two experi-
ments, with a new sample of subjects. Spe-
cifically, retention on delayed tests was re-
duced when the target trial was immediately
preceded by an initial trial that was either
the same as or opposite to the target trial.
As in the previous experiments, retention on
an immediate test was unaffected by the
occurrence of an initial trial.
The results of the second phase suggest
that requiring an initial run in each direction
is not a sufficient condition to eliminate in-
terference on an immediate test of retention.
Moreover, the between-phase comparison of
225
IS1 and IS1' and of ID1 and ID1' provided
direct support for the hypothesis that the
occurrence of a free-choice run as the ter-
minal event of a sequence of initial runs re-
sults in a significant improvement in target-
trial free-choice accuracy on an immediate
test. Although such between-phase compar-
isons must be interpreted with caution, sev-
eral factors suggest that such a comparison
may be less suspect in the present case than
in many others. First, the same subjects,
apparatus, and experimenter were used in
each phase. Second, although the two phases
were conducted sequentially, they were tem-
porally contiguous, and no training inter-
vened between phases.
Finally, the fact that 2PF produced greater
interference than either 1ST or IDT in the
second phase is interpretable readily in that
Condition 2PF involved two repetitions of
a forcing incompatible with the subsequent
target forcing. On the other hand, in both
1ST and IDT, animals received one com-
patible and one incompatible initial forcing.
The greater interference in 1ST than in IDT
replicates the finding from the first two ex-
periments that IS1 results in greater inter-
ference than ID1. The latter finding supports
further the notion that the initial runs es-
tablish independent memories and that the
degree to which a conflicting memory com-
petes successfully with the target memory
is dependent upon interfering memory re-
cency.
General Discussion
The present experiments have demon-
strated that presentation of an initial trial
immediately prior to the target trial reduces
accuracy significantly on delayed retention
tests but not on an immediate retention test.
Moreover, this effect was obtained whether
the initial trial was identical or opposite to
the target trial. Both the interval between
completion of the initial trial and initiation
of the target trial and the number of repe-
titions of the initial trial markedly influ-
enced the interference effect. Specifically,
amount of interference decreased with in-
creases in either intertrial interval or number
of repetitions of the initial trial.
Two of these findings may be accounted
226 DOUGLAS S. GRANT
for readily. The finding that retention was
reduced by an initial trial that was either
identical or opposite to the target trial sug-
gests that the initial trial was not remem-
bered as a unitary event. Since both identical
and opposite initial trials contained as a
component a run that would be expected to
interfere with retention of the target run, the
view that the memory of the initial trial is
composed of two at least partially separable
elements leads to the expectation of reduced
target trial accuracy given either type of ini-
tial trial. The finding that increasing the in-
terval between the initial and the target
trials reduced interference may be inter-
preted as an instance of forgetting. That is
to say, this finding follows directly from the
widely held assumption that memories con-
trol performance less strongly as time since
establishment (or activation) increases.
The rather counterintuitive finding that
repetition of the initial trial led to a decrease
in interference may prove more theoretically
intractable. One possible account might view
the organism in a short-term memory ex-
periment as discriminating among active
memories at the time of testing (see D'-
Amato, 1973; Grant, 1980, for such a view
of the monkey and rat, respectively). Ac-
cording to this view, repetition of the initial
trial may facilitate the rat's ability to dis-
criminate between memories established by
the initial runs and those established by the
target forcing. If repetition of the initial runs
are held to strengthen the memories estab-
lished by those runs, the rat may learn to
base the free-choice response on the weakest
memory in the system, a memory that should
correspond to the target forcing. This ac-
count encounters problems in the intratrial
preparation, however, in that repetition of
the initial forcing increases interference in
that preparation (Gordon et al., 1976; Rob-
erts, Note 1). Additional research is required
before a coherent account of the effects of
initial run repetition on interference in the
rat can be offered.
The final experiment addressed the issue
of which of the two procedural differences
between the intratrial and intertrial prepa-
rations is responsible for producing differ-
ential interference on an immediate test. The
data ruled out the notion that an initial run
to each arm of the maze is critical and
thereby provided indirect support for the im-
portance of the forced versus free nature of
the terminal initial run. Some direct evi-
dence favoring the hypothesis that the oc-
currence of a free choice attenuates inter-
ference on an immediate test was obtained
by comparing performance between phases.
The comparison revealed that retention was
more accurate on an immediate test when
the second initial run was free rather than
forced.
Perhaps the most reasonable interpreta-
tion of the importance of the occurrence of
a free-choice run in determining level of in-
terference on an immediate test is afforded
by a theoretical viewpoint in which a dis-
tinction between active and inactive mem-
ories is maintained (see Grant, 1980, in
press; Lewis, 1979; Spear, 1978, for further
discussions of the role of the active-inactive
distinction in contemporary theoretical ac-
counts of animal memory). According to this
view, the occurrence of a free-choice run as
the terminal event of a sequence of initial
runs increases the probability that the in-
terfering memory will be inactive at the time
of an immediate test of retention on the tar-
get trial. Since the contingencies of alter-
nation specify the irrelevance of specific re-
sponding on trial n for responding on trial
n + 1, it would be anticipated that the oc-
currence of the free-choice run would act as
a signal to discontinue active processing of
the prior forcing and free-choice memories.
Given a second, target trial and an imme-
diate test, the memory of the target-trial
forcing should be active since the free-choice
run has not been completed, while the mem-
ories of the initial runs should be inactive.
Hence, the animal should base the choice
response on the active memory of the target
forcing, and little interference on an im-
mediate test would be anticipated.
In the intratrial procedure, on the other
hand, the interfering event consists of an ini-
tial forcing(s) that is incompatible with the
target forcing. Since the animal has been
trained under contingencies that necessitate
retention of the direction of forced runs, the
rat should continue to process, and hence
keep active, the memory of the interfering
forcing. On a test of retention given imme-
 INTERFERENCE IN RAT SHORT-TERM MEMORY
diately following the target forcing, the rat
has two active memories, one representing
the interfering forcing and the other the tar-
get forcing. The rat must then discriminate
between the memories, presumably on the
basis of temporal attributes, that which is
the most recent. To the extent that such tem-
poral discrimination is an imperfect process,
the interfering memory should decrease ac-
curacy on an immediate test.
Finally, it may be noted that both intra-
trial and intertrial procedures have been em-
ployed to study proactive interference in the
pigeon. The paradigm employed has been
delayed matching to sample in which the
pigeon is required to choose between two
comparison stimuli the one that corresponds
to a previously presented sample stimulus.
Studies of intratrial interference involve the
successive presentation of two samples, one
corresponding to each comparison stimulus,
whereas studies of intertrial interference in-
volve the successive presentation of two com-
plete trials on which the role of correct and
incorrect comparison stimulus is reversed
between trials.
The data obtained with delayed matching
to sample in the pigeon contrast with those
presently obtained with the rat in two fun-
damental respects. First, repetition of the
interfering information increases interfer-
ence in both intertrial and intratrial prepa-
rations in the pigeon (Grant, 1975; Grant
& Roberts, 1973). On the other hand, the
present experiments have shown that repe-
tition decreases interference in delayed al-
ternation with the rat when the intertrial
procedure is used. Second, research with the
pigeon suggests that whether robust inter-
ference is obtained on an immediate test is
dependent upon the interval separating pre-
sentation of interfering and target infor-
mation (Roberts & Grant, 1976; Zentall
& Hogan, 1977). The present experiments
suggest that the critical factor is the occur-
rence or nonoccurrence of a free-choice run.
Additional research is required before it can
be determined whether these discrepant em-
pirical phenomena are a product of task or
species differences. Given rather limited un-
derstanding of the mechanisms of proactive
227
interference in rats and pigeons, it is possible
at present to do little more than note that
any complete theory of animal short-term
memory must be capable of accounting for
the influence of task and/or species vari-
ables.
Reference Note
1. Roberts, W. A. Proactive interference and short-
term memory in the rat. Unpublished manuscript,
1971.
References
D'Amato, M. R. Delayed matching and short-term
memory in monkeys. In G. H. Bower (Ed.), The psy-
chology of learning and motivation: Advances in re-
search and theory (Vol. 7). New York: Academic
Press, 1973.
Feldman, D. T. & Gordon, W. C. The alleviation of
short-term retention decrements with reactivation.
Learning and Motivation, 1979, 10, 198-210.
Gordon, W. C., Brennan, M. J., & Schlesinger, J. L.
The interaction of memories in the rat: Effects on
short-term retention performance. Learning and Mo-
tivation, 1976, 7, 406-417.
Gordon, W. C., & Feldman, D. T. Reactivation-induced
interference in a short-term retention paradigm.
Learning and Motivation, 1978, 9, 164-178.
Grant, D. S. Proactive interference in pigeon short-term
memory. Journal of Experimental Psychology: An-
imal Behavior Processes, 1975, /, 207-220.
Grant, D. S. Delayed alternation in the rat: Effect of
contextual stimuli on proactive interference. Learning
and Motivation, 1980, / / , 339-354.
Grant, D. S. Short-term memory in the pigeon. In
N. E. Spear & R. R. Miller (Eds.), Information pro-
cessing in animals: Memory mechanisms. Hillsdale,
N.J.: Erlbaum, in press.
Grant, D. S., & Roberts, W. A. Trace interaction in
pigeon short-term memory. Journal of Experimental
Psychology, 1973, 101, 21-29.
Lewis, D. J. Psychobiology of active and inactive mem-
ory. Psychological Bulletin, 1979, 86, 1054-1083.
Roberts W. A. Spaced repetition facilitates short-term
retention in the rat. Journal of Comparative and
Physiological Psychology, 1974, 86, 164-171.
Roberts, W. A., & Grant, D. S. Studies in short-term
memory in the pigeon using the delayed matching-to-
sample procedure. In D. L. Medin, W. A. Roberts,
& R. T. Davis (Eds.), Processes of animal memory.
Hillsdale, N.J.: Erlbaum, 1976.
Spear, N. E. The processing of memories: Forgetting
and retention. Hillsdale, N.J.: Erlbaum, 1978.
Zentall, T. R., & Hogan, D. E. Short-term proactive
inhibition in the pigeon. Learning and Motivation,
1977,5, 367-386.
Received June 30, 1980
Revision received December 17,1980 •