Zeelenberg, R., Boot,
Introduction
A robust false memory effect has been found repeatedly
in the DRM paradigm in previous studies.
Through the presentation of a list of semantic associates of a
non-presented lure, the episodic recognition of this lure at test has been
found to be as high or almost as high as recognition for studied items. This finding remains
even when warnings are given to participants.
Several explanations for these results were offered. The implicit associative response hypothesis
argues that the critical lure is brought to mind during study as a consequence
of it being related to the study words.
The spreading activation theories, on the other hand, argue that the each
studied word activates a series of associates either on a conscious or
unconscious level. Given that all list
items are related to the lure, the converging activation on the memory for nonpresented
lure engenders a false memory for it as a studied item. The contention of the authors is that the
better argument is that offered by global familiarity models such as MINERVA2. In these models, recognition at test is based
on comparison between the features of the test item and existing memory
traces. Greater similarity in this
comparison increases the likelihood an ‘old’ response will be given. These models argue that there need be no kind
of activation of the critical lure during study, but it’s recognition as ‘old’
at test is based entirely upon the similarity of the lure to the studied items
stored in memory. To offer further
evidence in support of these theories, the authors used a modified DRM paradigm
in which non-words were used as study and test items, eliminating the
likelihood of critical lure activation during study, and applied this under
both no-warning and warning conditions.
Experiment 1
Participants: 22 native Dutch-speaking students
participated for either class credit or
small
monetary reward
Materials: Sixteen 12-item lists were constructed of non-words
that were orthographic-
phonologically similar to a critical lure non-word and to each other. These words ranged in length from 4-8
letters, and were all orthographically legal and pronounceable. The critical lure was not necessarily most
similar to the rest of the list or served as a prototype. The list set was divided into two 8-list
sets, each shown to half of the participants.
The same test was administered to all.
The recognition test consisted of 64 items: 3 items from each of the
studied lists, the critical lure for each of the study list, 3 items from each
of the unstudied lists, the critical lure for each unstudied lists (total 64
items).
Procedure: The eight 12-itme study lists were shown in
an intermixed, random order
(total 96 items). Each
item was shown alone for 3 s on a computer screen, separated from the next by a
500 ms ISI. The study period was
immediately followed by the recognition test which was presented on the
computer, ‘old’ and ‘new’ responses collected via keyboard responses.
Results: A significant false memory effect was found
with participants recognizing the critical
lures for
studied lists 74.4% of the time vs. only 17.0% for unstudied list critical
lures. Also, the veridical memory effect
was significant. Participants recognized
86.4% of the words that had been studied, but only endorsed 14.6% of the list
items from unstudied lists as ‘old’.
Experiment 2
Participants:
19 native Dutch-speaking students participated for either class credit or
small
monetary reward.
Materials: Materials were identical to those used in
Experiment 1.
Procedure: Experiment 1’s procedure was replicated with
the addition of a warning. The false
memory
effect was explained to participants prior to the study phase, including
examples, and told not to make such mistakes.
Immediately before test, a warning was repeated.
Results: A significant false memory effect was found
with participants recognizing the critical
lures for
studied lists 58.6% of the time vs. only 12.5% for unstudied list critical
lures. Also, the veridical memory effect
was significant. Participants recognized
71.3% of the words that had been studied, but only endorsed 10.0% of the list
items from unstudied lists as ‘old’.
Though
not the intention of Experiment 2, the authors chose to look at how forewarning
participants affected the false memory effect by comparing the results of
Experiments 1 and 2. Calculating A’, a measure of memory discrimination,
showed no effect of warning. BD”, a measure of response
criterion, showed a dramatic increase for both critical lures and list items
when warnings were administered, showing an overall decreased likelihood to
make an ‘old’ response.
Discussion
The authors argue that their results cannot be
explained using implicit associative response accounts or spreading activation
theories because of the use of non-words.
Instead, these results offer solid support for global familiarity models. The authors find it unsatisfying to accept
that different mechanisms may be at play when the use of real words is used as
is put forth to support the spreading activation view. Global familiarity models can account for the
results found using non-words and words, as well as using very fast
presentation rates.
Important Legal Disclaimer: The preceding are articles we read together in the Lampinen Lab Spring 2006 false memory reading group. By clicking on the button next to the article you can see the summary of that article. The summary was prepared by the student presenting that article and it is of course the case that the views expressed in the summary do not necessarily represent the views of the reading group as a whole, Dr. Lampinen, the Lampinen Lab, Hugo's, the University of Arkansas, the Razorback Football or Basketball teams (although we're not sure about cross country), people living down the street from us, Bob Dylan, Jack Fate, our extended families, or anyone else for that matter except for the student who wrote the summary (and they don't necessarily believe what they wrote either).