Roediger
III, McDermott, and Goff (1977) called the effect of repeated testing on recall
and source memory the “paradoxical effects of repeated testing.” The present study uses the source monitoring
framework of Johnson and colleagues.
According to this framework, to remember the source of a memory, one
must use strategic judgment processes and heuristic judgments. Strategic judgment processes determine the
plausibility of an event and heuristic judgments are more automatic and rely on
general differences in memories.
Experiment 1
121
participants viewed 60 slides, half of which were black and white line drawings
of common objects with the name of the object printed across the bottom. The other half of the slides consisted of only
the name of the object across the bottom.
There were three types of stimulus pairs and within each pair, one
object was perceived and the other was imagined. The three types of stimulus pairs were:
1. The objects
physically resembled each other. (e.g. see a lollipop,
imagine a magnifying glass.)
2. The objects were
conceptually related to each other. (e.g. see a hairbrush, imagine a comb.)
3. The objects had
no physical or conceptual similarity. (e.g. see a sugar cube, imagine a candle.)
Participants
were told that the study was about how people make judgments between seen and
unseen objects. They were also told that
in some trials they would see the drawings of objects and in other trials, just
the name of the object and they were supposed to imagine the object that was
named. For both seen and unseen objects,
they were to think out loud of a normal function of the object. Six practice trials were given to familiarize
the participants and to reduce primacy effects.
The 60 slides were then presented one at a time for five seconds
each. Two filler trials followed the
last of the slides to reduce recency effects.
There
were two retrieval conditions. In both
of them, participants received a response sheet with 40 blank lines. In the forced recall position, participants
were told to recall as many of the items that were presented on the slides as
they could. If they got stuck before
reaching the 40th line, they were to stop and think some more and
recall more items. If they could not, then
they were to make an educated guess to fill up the 40 lines. In the forced source recall condition were
told to do the same thing but to also include whether the object was seen or
imagined, even on items that were guesses.
They were told they had 7 minutes to complete the task. After the 7 minutes, the sheet was removed
and they were told to think about the slides for five minutes. After those five minutes, the retrieval condition
was repeated. (Participants were given
the same instructions and to also include anything remembered on the last
memory test.) The response sheet was
again taken away and another five minute think period and another recall
test. After the final recall test,
participants in both conditions rated their confidence on each of the recall
items: 1=certain it was a picture,
2=fairly certain it was a picture, 3=not at all certain if it was a picture or imagined, 4=fairly certain it was imagined, 5=certain it was
imagined. A rating of X was included and
it meant that the participant was just guessing.
Results
Recalled imagined items increased from Test 1 to Test 2 and from Test
2 to Test 3 and was higher for participants in the forced recall condition compared to
the forced source recall. Recall was
also higher for items that were physically or conceptually similar than items
that had no similarity. The interaction
between specificity of memory test and test period were significant was
significant and showed that although initial recall was higher for the forced
recall than the forced source recall, retention did not differ for the two
tests on the last recall test. The
findings in Table 1 show that people are able to recall more with repeated
memory tests. The similarity between
seen and unseen objects helped participants to recall more objects with
physical or conceptual similarities. It
is possible that participants relied on their association with the similar
words to improve their recall.
Figure 1 shows that source errors increased across test periods and
more source errors were made for items with physical or conceptual similarities
than for those items with no similarities.
Significantly more source errors were made on Test 3 than Test 2 and
significantly more source errors were made on Test 2 than Test 1, which resulted
in more than twice as many source errors on the third test than the first. Figure 2 shows the effects of repeated memory
tasks on source errors for the third test.
More errors were made in the forced recall condition than in the forced
source recall and more errors occurred for physically or conceptually related
items. Also, on the third test, the
proportion of source errors (imagined items claimed as seen) was significantly
higher for items that had previously been recalled than those not recalled. This occurred in both conditions. Also, participants rated these falsely
claimed as seen objects as high in confidence that they had been seen.
Experiment 2
Experiment
2 was the same as Experiment 1 except the memory test was free recall as opposed
to force recall to control for guessing.
Results
The
results replicated those found in Experiment 1 in regard to recall rates and
source errors. Overall recall for
imagined items did not differ between experiments. However, people made significantly more
source errors on the forced recall than on the free recall. Also, the rate of increase of source errors
was more drastic in the forced recall than in the free recall. Participants were also much more likely to
report false claims with high confidence.
Experiment 3
The
procedure was the same as Experiments 1 and 2, with the exception that filler
stimuli was added. There were four
retrieval conditions, free recall and source recall and half of each condition took
three separate memory tests and the other half took one continuous memory test
for the same amount of time. Confidence
ratings were not made after the memory tests.
The memory tests were source recognition test where they were presented
with a list of items and they were to determine if the item had been seen,
imagined or was new. The test contained
124 items and had 84 old items in random order, mixed with 40 new items.
Results
Recall
again increased across test periods and was higher for free recall rather than
source recall. Again, there was higher
recall for physically or conceptually similar items. The number of retrieval efforts did not have
a significant effect on source errors, but the free recall tests produced more
source errors than the source recall test.
As in Experiments 1 and 2, more source errors were contributed to
similar items than control items.
Participants’
ability to determine old and new items was near ceiling but determining between
seen and unseen was poor (with rates of 30% and higher).
General Discussion
There
were three main findings with this study which were found in all three
experiments:
1. Recall increased and source
accuracy decreased with repeated memory tests.
2. Physically or conceptually similar
imagined items were more likely to be considered as seen.
3. Participants in the free recall
condition made more source errors than did those in the source recall condition.
This
poses a problem with false memories.
False memories can often be physically or conceptually similar to the
actual experience and, with repeated attempts to remember, a source error may
occur and people might believe their false memories actually occurred.