Eakin, D.K., Schreiber, T.A., Sergent-Marshall, S.
(2003). The Presence and Absence of
Memory Impairment as a Function of Warning and Misinformation
Accessibility. Journal of Experimental Psychology: Learning, Memory, and Cognition, 5,
813-825.
There are many reasons as to
why research on eyewitness memory is important. The main focus of the present article is to determine whether the
exposure to misinformation during the retention interval affects a persons
memory for details and if so, how. The
main causal factor discussed in this article is called “retrieval
blocking”. Retrieval blocking looks at
how original information retrieval is impaired because misinformation was
presented at test.
The general procedure the
researchers used was as follows. The
participants were shown a series of slides that depicted a maintenance man
going into an office, repairing a chair and steeling money and a calculator. To avoid floor effects the slide show was
shown twice. Four items in the slides
were labeled “critical” items. These critical items had two versions, one of
which was shown on the slides and one of which was used in the misled
condition. The items were: A jar of
instant coffee on a file cabinet (Maxwell House or Folgers), a fashion magazine
on an end table (Glamour or Vogue), a
can of soda on the desk from which the calculator is stolen ( Sunkist of 7-Up),
and the tool under which the man hides the calculator (screwdriver or
wrench). Participants would then
complete a booklet that contained a filler task of answering general knowledge
questions and then reading a narrative about the slides. Some participants would read a neutral
narrative (control) where the critical items were referred to in a generic
manner (e.g., tool) and some would hear a misleading narrative where the items
were referred to specifically (e.g., wrench).
This was followed by another set of general knowledge questions. After this, the MOT was administered. The MOT is the modified opposition test. The researchers used it to measure for
retrieval blocking. This is a modified
version of a test Lindsay used in the early 90’s. The MOT is a cued recall test combined with a list of rigid
instructions that warn the participant to ignore any information they received
in the narrative and answer the question asked bases solely on the information
seen in the slideshow. By using this
technique the researchers were able to control for such things as response
bias, social demand, and influence of source confusion.
The MOT consisted of four
cued recall questions (one for each critical item). Before taking the test, the participants in the mislead condition
were told that the narrative they read contained false information. The participants in the control condition
were simply told that some items would be referred to generically. All participants were told that if they read
a detail in the narrative for which they could answer one of the questions that
they should not consider it. All
participants also received a “hint” before each of the questions that indicated
the misinformation and warned against using it.
Experiment 1
The primary purpose of the
experiment was to see if the misinformation effect would occur if the
misleading information was only presented once. Generally when the MOT is administered, the misleading
information is presented several times.
The second goal of the experiment is to compare the performance of the
MOT to the MRT (modified recognition test).
The MRT works much like the MOT, however on the MRT the misleading item
is replaced as an answer choice with a semantically related item (e.g.,
screwdriver and hammer). This would
cause a decrease in the misinformation effect according to the researchers
because the absence of the misleading information would inhibit the
accessibility of it necessary for retrieval blocking.
The only difference in
Experiment 1 and the general procedures listed above was that the control and
misled items were manipulated within participants, thus, for each participant
two critical items were assigned to the control condition and two were assigned
to the misled condition. One group was
given the MOT and the other group was given the MRT. The researchers’ findings were as they predicted. There was a main effect of misinformation for
the MOT, but not for the MRT. These
data are important because this shows that the misinformation effect can be
obtained using the MOT when the misleading information is presented only once. This suggests that the MOT is useful for
exploring blocking effects.
Experiment 2
The purpose of Experiment 2
is to test whether blocking effects are greater when the misinformation is more
accessible. Instead of manipulating the
number of times the information is heard, Eakin, Schreiber &
Sergent-Marshall manipulate the placement of the information; either prior to
or following the event information. The
researchers also added the misled-reversed condition; a condition where the
narrative is read before the slide show is viewed. Eakin et al predicted that the misinformation effect would be the
lower for this condition than for the misled condition because the
accessibility of the misleading information is the weakest in this
condition.
The only difference in the
methods of Experiment 2 was that the misled-reversed participants first read
the narrative, answered the general knowledge questions, viewed the slide show,
answered the second set of general knowledge questions, and then took the
MOT. The results from this experiment
are found in Figure1. Again the researchers
found what they had predicted. The
lowest level of recall was for the misled condition, followed by the
misled-reversed condition, followed by the control condition. Statistical analyses show that the means for
all three conditions significantly differed from each other. These data follow the reasoning that
retrieval blocking is caused by accessibility to the misleading information.
Experiment 3
The purpose of Experiment 3
was to see how blocking effects can be reduced of eliminated by warning
participants about the misleading information immediately after it is
presented. Two new conditions were
added to the general procedure: a misled-immediate general warning condition in
which participants were simply warned that some of the information they
received was misleading and a misled-immediate specific warning condition, in
which the participants were told which bit of information had been
misleading. The researchers predicted
that the specific warning would cause a larger misinformation effect because
the misleading information would be cued when it was presented to the
participant as being false, thus, making it more accessible than the
information in the general warning condition.
The other result we could possibly see is that the specific would
actually elicit a lower misinformation effect by triggering some sort of
suppression mechanism.
The results of Experiment 3
can be seen in Figure 2. As in the
first two experiments, there was a significantly smaller misinformation effect
for the control condition than for the misled-warning at test only condition,
but, there was no significant difference between the control condition and the
two new conditions. These data show
that participants do use some type of suppression mechanism when they are
presented with a warning directly after exposure to the misleading information,
no matter how specific the warning. The
idea behind a suppression mechanism is that when a narrative is read, all
information, even misleading, is made more accessible. When the warning is given, participants try
to suppress that level of activation of either the entire narrative (contextual
suppression) or just the misleading information (item-specific suppression)
until it is low enough that they will not recall it during test time. The results of this study correspond more
with the idea of contextual suppression.
The major finding of Experiment 2 was that retrieval blocking effects
were eliminated apart from of the type of warning that was given.
Experiment 4A
Apart from replicating the
results from Experiment 3, Experiment 4A was done to determine if the
suppression hypothesis was strong enough to explain the data under which
suppression is difficult. In addition
to the four conditions in Experiment 3, there were four “high-accessibility”
conditions added. These high
accessibility conditions can be viewed in Figure 3. An interpolated test was included in the experiment to force the
participants to think about and identify the misleading information. The misleading information was heard twice
instead of once in the high accessibility condition. The researchers predicted the same types of results we saw in
Experiment 3 for the low accessibility condition. The experimenters also hypothesized that the high accessibility,
immediate warning conditions would most likely look like the results obtained
from the misled condition in which no warning was given.
The results for this
experiment can be seen in Figure 4. The
low accessibility list replicated Experiment 3, just as the researchers
suggested. These patterns were not the
same for the high accessibility conditions.
Not only was a bigger misinformation effect found for the misled
warning-at-test-only condition when compared with the control condition, the
was also a higher level of the misinformation effect in the immediate warning conditions as well, showing that when
the misinformation is highly accessible it does not matter when the warning is
given. These data suggest that while a
suppression mechanism is helpful when accessibility is low, it does not seem to
make a difference when accessibility is high.
Experiment 4B
There was one interpretive
problem in Experiment 4A. The
researchers interpreted the results in Experiment 4A as proof that warnings are
less effective at reducing the misinformation effect when the misleading
information is more accessible. It can
be argued thought that this decrease in effectiveness is not caused by an
increased in accessibility to misleading information, but to a decreased accessibility
to the original information. This
problem can be best seen in Figure 5.
The interpolated test caused the time interval between the slide show
and the test to be about 10 minutes longer for the high accessibility
conditions as opposed to the low accessibility conditions. To remedy this, the researchers administered
the interpolated test to all conditions.
The results for Experiment 4B
can be seen in Figure 6. This study
replicated the findings from Experiment 4A.
This suggests that the time that elapses between the slide show and the
test does not seem to show an affect, meaning that the original interpretation
made by the researchers seems to be the most accurate.
General Discussion
There is still a lot more
that needs to be researched in the area of the misinformation effect. The present study shows that the idea of
retrieval blocking is a valid way to explain this phenomenon. Some areas that need more research are the
areas of the suppression hypothesis and the timing of the warning. Looking at the article in practical terms
does not make one feel hopeful. These
patterns of results show that eyewitnesses could fall prey to the
misinformation effect even with multiple warnings.