YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
No. of pages: 16
DTD 4.3.1 / SPS
Journal of Memory and Language xxx (2003) xxx–xxx
Journal of
Memory and
Language
www.elsevier.com/locate/jml
5
D. Stephen Lindsay,a,* Bem P. Allen,b Jason C.K. Chan,c and Leora C. Dahla
3
a
6
7
8
b
9
PR
OO
F
4
Eyewitness suggestibility and source similarity: Intrusions
of details from one event into memory reports
of another eventq
2
Department of Psychology, University of Victoria, Vic., Canada
Department of Psychology, Western Illinois University, IL, USA
c
Department of Psychology, Washington University, USA
ED
Received 30 May 2003; revision received 19 August 2003
10 Abstract
RE
CT
We explored the effect of the degree of conceptual similarity between a witnessed event and an extra-event narrative
on eyewitness suggestibility. Experiments 1A and 1B replicated Allen and LindsayÕs (1998) finding that subjects
sometimes intrude details from a narrative description of one event into their reports of a different visual event. Those
experiments also showed that intrusion rates were even higher when the narrative described the visual event itself.
Experiment 2 replicated those findings, but found no more intrusions from a thematically similar versus dissimilar
narrative. In Experiment 3 we disguised the relationship between the narrative and visual event, and obtained more
intrusions from a thematically similar than dissimilar narrative. In Experiment 4 we obtained a thematic similarity effect
when the relationship between narrative and visual event was disguised, but none when it was not. Results are discussed
from the perspective of the source-monitoring framework.
Ó 2003 Published by Elsevier Inc.
CO
R
11
12
13
14
15
16
17
18
19
20
21 Keywords: Memory; Eyewitness suggestibility; Misinformation effect; Source monitoring
22 Introduction
In LoftusÕs eyewitness misinformation paradigm,
participants view a visual event, receive verbal information about that event that includes misleading suggestions, and are later tested on their memory of the
event. Under a wide variety of conditions, participants
sometimes err by reporting details suggested in the ver-
UN
23
24
25
26
27
28
q
This research was supported by a research grant from the
Natural Sciences and Engineering Research Council (NSERC)
of Canada to the first author, and by a NSERC postgraduate
scholarship to the fourth author. We thank Larry L. Jacoby
and two anonymous reviewers for helpful critiques of an earlier
version of the manuscript.
*
Corresponding author. Fax: 1-250-721-9829.
E-mail address: [email protected] (D. Stephen Lindsay).
0749-596X/$ - see front matter Ó 2003 Published by Elsevier Inc.
doi:10.1016/j.jml.2003.08.007
bal information on tests of their memories of the witnessed event. For example, participants in a study by
Loftus, Miller, and Burns (1978) viewed a slide show
depicting an accident in which a car hit a pedestrian; for
some participants, one slide showed the car at an intersection marked by a stop sign but a subsequent misleading question implied that the intersection had been
marked by a yield sign. On a later memory test, such
participants often falsely indicated that they had seen a
yield sign in the slides.
In the 1980s, research on the misinformation effect
emphasized the question of whether or not suggested
details impair ability to remember the corresponding
event details (memory impairment) (e.g., McCloskey &
Zaragoza, 1985). From this perspective, the misinformation effect can be construed as an extension of earlier
research on interference phenomena in verbal learning
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
be restricted to things seen in that event. The key finding
was that participants sometimes intruded details from
the narrative into their reports of the slides. Presumably,
these errors occurred because (a) the test questions were
good cues to memories of the narrative (e.g., due to
semantic overlap between questions and suggestions)
and (b) participants sometimes accepted memories that
popped to mind at test as accurate answers (e.g., because
those memories were plausible answers).
One reason for interest in the misinformation effect is
that it may generalize to real-world cases in which eyewitnesses are asked questions about a consequential
event. To the extent such effects do generalize, the Allen
and Lindsay (1998) findings raise the possibility that
eyewitnessesÕ testimony may be compromised not only
by suggestions regarding the witnessed event but also by
memories of other events. Such information might come
from a variety of sources (e.g., TV, books, other personal or vicarious experiences). It is also worth noting
that deliberate uses of extra-event information are unlikely when that information concerns a different event
(because, unlike in standard misinformation paradigms,
there is no reason for subjects to assume that the narrative is a source of accurate information about the tobe-recalled event).
The current paper has two major aims. First, we
sought to replicate the findings of the Allen and Lindsay
(1998) experiment (which is the only extant demonstration of intrusions from a narrative description of one
event into eyewitness reports of another event) with a
variety of materials. Second, we assessed the role of the
degree of conceptual similarity between the witnessed
event and the narrative in modulating the rate of such
intrusions. In the Allen and Lindsay experiment, the two
sources were thematically similar (i.e., both concerned an
interaction between a professor and a student). The SM
framework implies that, all else being equal, intrusions
should increase as a function of the conceptual similarity
between the two sources (a prediction also made by
earlier theories; e.g., Abra, 1972; McGeoch & McGeoch,
1937). Interestingly, however, prior studies on the effects
of various kinds of source-similarity manipulations on
the misinformation effect (reviewed in more detail below)
have produced inconsistent results. Across the five experiments reported here, we explored conditions under
which the degree of conceptual similarity between the
target event and narrative does versus does not affect the
rate of intrusions. Before reporting the experiments, we
describe the theoretical context in which we pursued this
work and summarize prior studies of the relationship
between source similarity and suggestibility.
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
The source-monitoring framework
153
The source-monitoring (SM) framework holds that
all experiences of remembering involve inferential pro-
154
155
CO
R
RE
CT
ED
studies (e.g., Keppel & Underwood, 1962; McGeoch,
1936; McGovern, 1964; Postman & Stark, 1969; for a
more recent treatment of the memory-impairment issue
in the context of the misinformation effect, see Chandler,
Gargano, & Holt, 2001).
In the 1990s, the interest of researchers exploring the
misinformation effect shifted to the question of whether
participants believe that they are remembering something they witnessed when they report a suggested detail,
as opposed to knowingly relying on the verbal information (either because they assume it is a reliable source
of information about the event or because they think the
experimenter wants them to report details from the
verbal information). The issue of whether subjects
sometimes confuse memories of suggestions as memories
of the witnessed event can be cast as an extension of
earlier verbal-learning research on list differentiation
(e.g., Abra, 1972).
It is clear that participants in misinformation studies
sometimes knowingly use memories of extra-event information when answering questions about a target
event, but it is also clear that participants sometimes
falsely believe they remember witnessing details that were
merely suggested to them. For example, when conditions
are such that it is easy for participants to differentiate
between memories of the event and memories of the
verbal information, informing participants immediately
before the test that the verbal information was false and
should not be reported eliminates the suggestibility effect,
but when conditions make it difficult for participants to
differentiate between memories from the two sources,
such instructions attenuate but do not eliminate the
suggestibility effect (Lindsay, 1990; Lindsay, Gonzales, &
Eso, 1995; see also Holliday & Hayes, 2000, 2001, 2002;
Roediger III, Jacoby, & McDermott, 1996).
In most versions of the misinformation paradigm, the
misleading suggestions are embedded in verbal information about the witnessed event. For example, participants might read or hear a narrative summary of the
event that includes misleading suggestions, or they might
answer questions about the event that include inaccurate
suppositions (e.g., ‘‘How fast was the car going when it
passed the barn?’’ when there was no barn). Allen and
Lindsay (1998), however, found that participants
sometimes intruded details from a verbal description of
one event into their memory reports of a different event.
In that study, participants viewed a slide show of one
event (e.g., a male professor conversing with a female
student in a lecture hall) and read a narrative describing
a different event (e.g., a female professor conversing with
a male student in an office). The two events were unambiguously different, but each included certain common details (e.g., a can of pop was shown in the slides,
and a different can of pop was mentioned in the narrative). Participants were tested on their memory of the
visual event, with instructions that their reports should
UN
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
PR
OO
F
2
No. of pages: 16
DTD 4.3.1 / SPS
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
No. of pages: 16
DTD 4.3.1 / SPS
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
Eyewitness suggestibility and the SM framework
245
The SM framework has been developed and explored
using a variety of research procedures, including variants of the misinformation paradigm. Here we briefly
summarize prior studies of eyewitness suggestibility that
were inspired by (and informed further developments of)
the SM framework.
Lindsay and Johnson (1989b) demonstrated a standard suggestibility effect in a procedure in which exposure to verbal suggestions preceded (rather than
followed) witnessing the visual event (see also Abeles &
Morton, 1999; Holliday & Hayes, 2002). Those findings
are consistent with the idea that misinformation effects
can reflect confusions between different memory sources
(as opposed to reflecting a mechanism by which new
information updates memory representations of prior
information).
Lindsay and Johnson (1989a) found that a suggestibility effect obtained with a yes/no recognition test was
absent among participants tested on a SM test that required them to specify the source (visual event, narra-
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
PR
OO
F
Source-monitoring accuracy is also affected by the
appropriateness and stringency of the decision-making
processes and criteria used. Under many conditions,
people largely disregard the sources of the thoughts and
images that come to mind (because source discrimination is not relevant to the task at hand; e.g., if the goal is
to solve a problem one may not attend to the memorial
sources of the potential solutions that come to mind).
Under other conditions, people make source attributions quickly and automatically, and under yet others
they set stringent criteria and scrutinize memory information systematically before making a source attribution. All else being equal, SM accuracy increases with
the stringency and appropriateness of the criteria and
processes used (although errors may sometimes occur
even with stringent and appropriate criteria, e.g., if the
two sources are highly similar).
As in many areas of cognitive psychology, the various factors hypothesized to influence SM have usually
been investigated cetaris paribus (Palmer, 1999, cited in
Vecera, Vogel, & Woodman, 2002). Thus at this stage in
its development the SM framework makes few predictions regarding interactions between the parameters
thought to affect source monitoring (e.g., dimensions of
similarity [visual, auditory, conceptual, etc.], expectations, aspects of source [where, who, how, when?], the
precision with which sources are identified, and the
kinds and stringency of source-monitoring criteria).
Manipulating a particular dimension of source similarity, for example, may have no effect if, at test, participants make their source attributions on other
dimensions or do not attend to source at all (see Marsh
& Hicks, 1998).
CO
R
RE
CT
ED
cesses by which people attribute thoughts, images, and
feelings to particular episodes in their personal past experience (for reviews, see Johnson, Hashtroudi, &
Lindsay, 1993; Johnson & Raye, 2000; for closely related
views, see Jacoby, Kelley, & Dywan, 1989a; Whittlesea,
2003). SM attributions are thought to be based on the
qualitative and quantitative characteristics of the content of accessed memory information, in the context of
the testing situation. The notion is that memory information is attributed to particular sources in a way that is
analogous to how current perceptual information is attributed to particular sources. For example, if a friend
calls you on the phone, you may immediately recognize
the person by his or her voice (perhaps also influenced
by other sorts of knowledge, such as your expectancy
that so-and-so is likely to call at a particular time of
day). So too, if you recollect an utterance you may attribute it to a past experience of hearing someone speak,
and more specifically you may attribute the statement to
your friend Jean because the accessed memory information includes information about the sound of JeanÕs
voice and/or because the semantic content of the utterance is in some way characteristic of Jean.
Source attributions can be made at a wide range of
degrees of precision (e.g., from ‘‘I once read somewhere...’’ to ‘‘Yesterday afternoon at the checkout
counter at Safeway I noticed the National Enquirer. . .’’),
depending partly on the remembererÕs current goals.
Like ongoing perceptual attributions, most memorysource attributions are made quickly and without conscious deliberation, but people sometimes have difficulty
identifying source at the desired level of specificity (in
which case they may consciously strive to retrieve additional source-specifying information and use analytic
strategies to identify the source of memory information).
Source attributions are usually accurate, but sometimes they are faulty (e.g., you may think that Jean said
something you had actually heard from Francis; that
something occurred yesterday when it was really the day
before; that you turned the oven off when you had
merely thought about doing so; that you came up with
an idea that you actually read, etc.). All else being equal,
memory-source attributions are easier (and errors less
likely) when the accessed memory information is uniquely characteristic of its true source. For example, if
you have two friends with similar voices, you may be
liable to mistake memories of an utterance by one as
having been made by the other (just as you might mistake their voices on the phone). So too, as the amount of
accessed memory information declines its source diagnosticity typically declines as well, such that SM tends to
be more difficult if the event was not closely attended to
during encoding, if there is a long delay between encoding and test, or if memory test conditions interfere
with retrieval (akin to not recognizing a voice on the
phone when the connection is bad).
UN
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
3
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
means were in the predicted direction, with a 37% difference in reports of suggestions on misled vs. control
items in the same-context condition as opposed to only a
22% difference in the different-context condition). In two
studies with bilingual participants, Shaw III, Garcia,
and Robles (1997) reported null effects of same versus
different language during the event and misinformation
phases on performance on a subsequent test (2AFC or
cued recall). More recently, Mitchell and Zaragoza
(2001) reported two experiments in which they manipulated the extent to which the misleading narrative recapitulated the event. Suggestibility effects on a SM test
were reliable and of the same magnitude regardless of
whether the narrative was a detailed and coherent
summary of the event or a relatively disjointed and brief
series of statements ordered differently from the sequence in the event itself.
Why do source-similarity manipulations modulate
the size of the misinformation effect under some conditions but not others? One possibility is that the inconsistency has to do with variations in the extent to which
deliberate uses of extra-event information contribute to
apparent suggestibility effects. As noted above, participants sometimes select or report items that they know
they recollect encountering in the extra-event material,
either because they assume that material is reliable or
because they think the experimenter wanted them to rely
on that material. Aware uses of extra-event information
are especially likely when that information is apprehended as being about the to-be remembered event and
is presented shortly before the test and/or processed in
ways that support good memory for the suggestions, and
when the test instructions do not warn participants
against relying on the extra-event information. To the
extent that reports of suggested items reflect aware or
deliberate uses of extra-event information, there is no
reason to expect that source-similarity manipulations
would affect the rate of such errors.
A related point is that standard 2AFC tests (like that
used by Bonto & Payne, 1991) may encourage participants to use a familiarity criterion for response selection.
Because most test items pair a witnessed detail with a
novel foil, participants may slip into a tendency to select
whichever response ‘‘pops out’’ as familiar (unless they
are explicitly warned against doing so) (see Lindsay &
Johnson, 1989b). Here again, to the extent that participants base their responses on familiarity without regard
to source there is little reason to believe that sourcesimilarity manipulations would be effective.
It is also likely that different sorts of source-similarity
manipulations have differential effectiveness. For example, participants in the Shaw et al. (1997) studies were
fluent Spanish–English bilinguals living in El Paso,
Texas; such individuals are likely to have considerable
practice at abstracting meaning away from its linguistic
medium (and in any case the witnessed event was pre-
CO
R
RE
CT
ED
tive, both, or neither) of each test item (see also Zaragoza & Koshmider, 1989). Lindsay and Johnson interpreted these findings as evidence that the SM test
encouraged participants to attend to source-specifying
memory information, thereby enhancing their ability to
differentiate between memories of the target visual event
versus memories of the verbal information. Converging
evidence for that conclusion came from Zaragoza and
Lane (1994, Experiment 2), in which suggestibility effects
were reduced but not eliminated by a SM test as opposed to yes/no test (see also Chambers & Zaragoza,
2002; Hekkanen & McEvoy, 2002; for an interesting
exception, see Hicks & Marsh, 2001).
Zaragoza and Lane (1994) found that presenting
suggestions in ways that encourage participants to encode them elaboratively and in integration with event
memories increased reports of suggestions on a subsequent SM test (see also Mitchell & Zaragoza, 1996).
Lindsay (1990) adapted JacobyÕs (e.g., Jacoby, Woloshyn, & Kelley, 1989b) opposition procedure, by correctly informing participants before a cued recall test
that any potential answer that had been mentioned in
the narrative was a misleading suggestions and hence
should not be reported: Consistent with the SM framework, under conditions that made it easy for participants to differentiate memories of the narrative from
memories of the event, these instructions enabled participants to avoid reporting suggestions, but under
conditions that made SM difficult participants often
intruded suggestions despite the opposition instructions
(see also Holliday & Hayes, 2000, 2001, 2002; Lindsay et
al., 1995; Roediger et al., 1996). Eberman and McKelvie
(2002) found higher rates of reports of suggestions on a
SM test among participants with high imagery than
among those with low imagery. Frost, Ingraham, and
Wilson (2002) found that, as predicted by the SM
framework, source accuracy declined more steeply than
recognition memory over a 1-week delay.
A key premise of the SM framework is that, all else
being equal, increasing the similarity between memories
of suggestions and memories of the witnessed event will
increase false reports of suggestions. Several studies cited above support that conclusion, but there are also
published reports of null effects of source-similarity
manipulations on suggestibility.1 Bonto and Payne
(1991) found no significant effect of same versus different
environmental context during the event and postevent
information on performance on the standard Loftus 2alternative forced-choice (2AFC) test (although the
UN
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
PR
OO
F
4
No. of pages: 16
DTD 4.3.1 / SPS
1
Null effects of source-similarity manipulations have also
been reported in several experiments using McCloskey and
ZaragozaÕs (1985) modified test, but we do not discuss such
findings here because the modified test precludes subjects from
reporting suggestions and hence is likely to be insensitive to
source-similarity manipulations (Lindsay & Johnson, 2001).
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
No. of pages: 16
DTD 4.3.1 / SPS
391 Experiments 1A and 1B
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
Method
450
Participants
There were 303 participants in Experiment 1A; as
explained below, 15 of these were dropped due to poor
performance on easy filler items on the memory test, so
analyses are based on the remaining 288 (57% women).
Most (56%) were college/university students who earned
credit toward a course at a college or university for their
participation, but others were people of a variety of ages
who spontaneously volunteered to participate (age range
from 14 to 56 years, M ¼ 24:81; SD ¼ 9:80). Most participants (85%) were from the United States, but countries around the world were represented. Participants
logged on to the experiment through various research
sites (e.g., the American Psychological Society site).
The participants in Experiment 1B were 146 undergraduate students at Western Illinois University. Two
participants were dropped due to poor performance on
easy filler questions on the memory test, so analyses
were based on the remaining 144 participants (64% female, age range from 17 to 23 years, M ¼ 19:03;
SD ¼ 1:30). They received credit points in an introductory psychology class in exchange for participating.
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
Materials and procedure
The materials and procedures in the two experiments
were essentially the same, but Experiment 1A was
conducted on the web whereas Experiment 1B was
473
474
475
476
CO
R
RE
CT
Experiments 1A and 1B had three major aims. First,
we sought to replicate, with a new set of materials, Allen
and LindsayÕs (1998) finding that participants sometimes
intrude details from a narrative describing one event
into their reports of another witnessed event. Second, we
went beyond the Allen and Lindsay (1998) procedure by
decreasing the thematic similarity between the two
events: Whereas in Allen and LindsayÕs study the witnessed event and narrative both involved student-professor conversations, in the present experiments the
witnessed event involved searching a house for a misplaced object and the narrative involved housecleaning.
Third, we compared the rates of intrusions from a different-event narrative versus from a narrative describing
the visual event itself.
From the perspective of the SM framework, memories of a narrative description of a witnessed event
should be more similar to memories of that event itself
than are memories of a narrative description of a different event. This is because semantic content is an important cue to source (e.g., Lindsay, Johnson, & Kwon,
1991, Experiment 3). Information that is apprehended
and encoded as being about the witnessed event should,
from this perspective, give rise to memories that are both
(a) likely to come to mind during attempts to recollect
relevant aspects of the to-be-remembered event and (b)
likely to be accepted as memories of the event itself.
Thus we predicted that the misinformation effect would
be larger in the same-event narrative condition than in
the different-event narrative condition.
When the misleading narrative describes the target
event, there is a possibility that participants will know-
UN
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
ingly use memories of the narrative as a source of answers at test. We sought to reduce that possibility by
instructing participants to base their answers solely on
what they remembered seeing in the target event. To
provide converging evidence as to the extent to which
participants knowingly relied on the narrative as a
source of answers, Experiments 1A and 1B also included
conditions in which the test instructions warned participants that the narrative included details not present in
the slides. This is a relatively weak warning (much
weaker than the opposition instructions, which tell
participants that any relevant detail remembered from
the narrative is wrong and should not be used as a basis
for responding), but we reasoned that if participants in
the non-warned conditions were knowingly using
memories of the narrative as a basis for answers then
this warning would reduce such aware uses of the extraevent information.
The experiments also compared a Web-based version
of the procedure (Experiment 1A) with a laboratory
procedure (Experiment 1B). Prior studies indicate that
Web research can yield results comparable to ‘‘live’’
research (Allen, 1999; Birnbaum, 2000; McGraw, Tew,
& Williams, 2000), but to our knowledge no previous
study has compared Web-based and laboratory versions
of a misinformation procedure.
sented silently; only the instructions that preceded and
followed it were in one language or another), so the
similarity of memories of the event versus the extraevent information may not have differed much as a
function of same versus different language. Likewise, the
manipulation of the extent to which the narrative recapitulated the event in the Mitchell and Zaragoza (2001)
studies may have had little effect on the extent to which
memories of particular details from the narrative were
similar to memories of details from the event: In both
cases, participants likely apprehended the narrative as
being about the event. In the absence of a validated
model of the psychological construct of similarity, these
are merely speculations, but in our judgment they are
plausible ones.
With these considerations in mind, in the experiments
below we used strong similarity manipulations, test instructions that clearly specified that responses were to be
based solely on memory for the visual event, and cued
recall tests rather than 2AFC.
PR
OO
F
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
5
ED
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
scribed as being the same as that in the target event, and
the order in which the object clusters were encountered
differed from that in the target event.
Participants were admonished on the narrative page
that if they returned to the slide show their responses
would be invalidated and their time, as well as that of
the experimenter, would be wasted (a similar warning
regarding returning to the stories was printed at the
beginning of the next page, described below). After the
participants read the stories, they clicked a link to a
questionnaire page in CGI format that asked for consent
to use their data and asked them to report age, sex, and
place of residence.
Next, a series of filler questions focused on the filler
stories to support the perception that all of the stories
were important. Finally, a series of questions about the
target-event slide show was preceded by a plea that
participants answer the questions solely on the basis of
their memories of the slides presented in the first phase
of the experiment. They were also asked not to leave any
question unanswered (although 38.5% of participants in
Experiment 1A and 26.0% of those in Experiment 1B left
unanswered one or more questions for which they had
received suggestions). For approximately half of the
participants, the test instructions warned that some of
the details mentioned in the second story were NOT in
the slides, and ‘‘you cannot assume that just because
something was mentioned in the story it is a correct
answer to the test questions about the slide show.’’ The
remaining participants did not receive this warning.
Participants were assigned to the ‘‘warning’’ and ‘‘no
warning’’ conditions by changing the link from the
narrative page to the final questionnaire every few days.
There were two critical cued-recall questions regarding each object cluster (one for each ‘‘marker’’ object), with one question asking about a control item and
the other asking about a suggested item (e.g., ‘‘What was
near the outer, rounded edge of the hand mirror that
was on the dressing table top?’’ was a control question
for participants assigned to Lure Set A [who had read no
suggestion regarding something close to the mirror] and
an experimental question for participants assigned to
Lure Set B [who had read a suggestion that a cotton
swab was close to the mirror]). Intermingled among the
critical questions were four ‘‘easy’’ questions regarding
details in the slide show that anyone who had viewed the
slides would likely remember. These questions were included to establish that participants had the slide show
in mind when answering questions about it. As mentioned earlier, participants who made more than one
error on the four easy questions (15 participants in Experiment 1A [5%] and 2 in Experiment 1B [1%]) were
excluded from all analyses.
The slides can be seen at http://www.wiu.edu/users/
mfbpa/slideshow.html. The stories can be viewed at the
same URL, substituting the following for ‘‘slideshow:’’
CO
R
RE
CT
ED
conducted in a laboratory with groups of 2–11 participants and a ‘‘live’’ experimenter. For Experiment 1B, the
web pages used in Experiment 1A were printed (with the
slides in color, as on the web site) and distributed to
participants in booklet form (with pages in the same
order as that of the web pages), and participants wrote
in the buttons and dialog boxes with a pencil. Because
the procedures were otherwise identical, below we provide details only of Experiment 1A.
The first web page provided a brief description of the
study and informed consent information, followed by a
text describing the target event illustrated with color
slides. The target event involved a man searching his
house to find a misplaced gold coin. As his search progressed through different rooms, he encountered six
clusters of from 9 to 14 objects: tools (including a saw
and plumberÕs wrench); beauty products (tweezers and
mirror); small ceramic animals (mouse and rabbit); a
place setting (plate and mug); home office supplies
(magic marker and stapler); and home hardware (bolt
and nail). Each cluster was depicted in a separate slide,
and the objects in parentheses were mentioned in the
accompanying narrative (e.g., ‘‘On the dressing table top
he found an assortment of hair-care and make-up items.
There were some tweezers in the middle of the items
arrayed on the table top. On the far left edge of these
items there was a large hand mirror’’). We refer to the
objects mentioned in the text (e.g., ‘‘mirror’’) as ‘‘marker
objects,’’ because objects not actually present in the
slides—suggested lures—were ‘‘placed’’ next to them in a
subsequent misleading narrative (e.g., ‘‘He picked up a
cotton swab that was next to a mirror, then tossed it
back onto the counter-top’’). The lures for each cluster
were as follows (List-A lure first): pliers, screwdriver
(tools); cotton ball, cotton swab (beauty products);
squirrel, raccoon (small ceramic animals); salt shaker,
butter dish (place setting); eraser, pencil (home office
supplies); drill bit, nut (hardware).
After reading the event narrative and looking at the
slides, participants clicked a link to a page that contained three stories: beginning and ending filler stories
and a second-position misleading narrative. Participants
were assigned to one of two misleading narratives by
alternating the link to them every few days. One narrative was the same as the event, in that it described the
same ‘‘gold coin’’ scenario as in the initial event, with the
same character searching the same house and encountering the six clusters of objects (with one marker object
and one suggested object per cluster). The other narrative was different from the event: Although there were
still certain parallels between the target event and the
narrative, in that both involved a protagonist moving
through a house and encountering various clusters of
objects, in the different-event condition the protagonist
was a teenage girl rather than a man, her goal was to
complete house-cleaning chores, the house was not de-
UN
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
PR
OO
F
6
No. of pages: 16
DTD 4.3.1 / SPS
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
No. of pages: 16
DTD 4.3.1 / SPS
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
592 Results
‘‘Easy’’ questions
For the ‘‘easy’’ questions, mean proportion accurate
was near ceiling (Ms ¼ :974 and .897 in Experiments 1A
and 1B, respectively), with no reliable difference across
counterbalancing and warning/no-warning conditions
(all F s < 1).
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
Probability of false recall
In each experiment, the proportion of the six critical
items on which participants reported suggested details
was analyzed in a 2 (suggested vs. control) 2 (sameevent vs. different-event narrative) 2 (warning vs. no
warning) 2 (counterbalancing A vs. B) analysis of
variance (ANOVA), in which the first variable was a
repeated measure. Effect size was indexed with partial
eta squared (es; .01 or less is a small effect, .06 a moderate effect, and .14 and above a large effect). Descriptive
data are depicted in Fig. 1.
In each experiment, there was a reliable suggestibility
effect, as revealed by a significant effect of item type
(suggested vs. control): in Experiment 1A, F ð1; 280Þ ¼
97:44; MSe ¼ :01305; p < :001; es ¼ :258, and in Experiment 1B, F ð1; 136Þ ¼ 48:12; MSe ¼ :01747; p <
:001; es ¼ :261. There were also reliable effects of the
between-subjects manipulation of narrative type: in
Experiment 1A, F ð1; 280Þ ¼ 19:00; MSe ¼ :01242; p <
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
Discussion
634
Results of the internet and laboratory versions of the
procedure were nearly identical, supporting the viability
of web-based experiments on the misinformation effect.
Moreover, the Allen and Lindsay (1998) finding was
replicated and extended with these new materials and
procedures, in that participants in the different-event
condition sometimes intruded details from their memories of the house-cleaning narrative into their reports of
the gold-coin-search slides. The finding that a thematically distinct story can give rise to false reports dramatically extends the range of potential sources of
contamination in eyewitness reports, even relative to the
Allen and Lindsay study.
As predicted, the rate of false reports was substantially and significantly higher when suggestions were
embedded in a narrative description of the witnessed
event than when the same suggestions were presented in
a narrative description of a different event. This finding
is consistent with the SM framework: Memories of the
narrative would be expected to be more similar to
memories of the visual event in the same-event narrative
condition than in the different-event condition, and
hence participants in the same-event condition would be
more likely to retrieve memories from the narrative at
test and more likely to accept them as memories of the
target event.
It is possible that participants in the same-event
condition sometimes based responses on memories that
they knew came only from the narrative (despite the test
instructions to base reports only on the slides). Because
there would be little reason for those in the differentevent condition to base reports on memories they knew
came only from the story (given that the story was
clearly about a different event), such aware uses of storyonly memories in the same-event condition may have
contributed to the difference in intrusion rates in the two
conditions. As evidence against that possibility, the null
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
UN
CO
R
RE
CT
ED
593
594
595
596
597
598
:001; es ¼ :064, and in Experiment 1B, F ð1; 136Þ ¼
18:71; MSe ¼ :01805; p < :001; es ¼ :121. Most importantly, there was a reliable Item Type Narrative
Type interaction in each experiment: in Experiment 1A,
F ð1; 280Þ ¼ 16:04; MSe ¼ :01305; p < :001; es ¼ :054,
and in Experiment 1B, F ð1; 136Þ ¼ 10:19; MSe ¼
:01747; p < :001; es ¼ :07. No other main effects or
interactions were significant (all Fs < 3:5; all ps > :069).
Follow-up tests showed reliable suggestibility effects in
both the same-event condition (Experiment 1A:
F ð1; 143Þ ¼ 68:914; MSe ¼ :01823; p < :001; es ¼ :325;
Experiment 1B: F ð1; 71Þ ¼ 33:22; MSe ¼ :02697; p <
:001; es ¼ :327) and the different-event condition (Experiment 1A: F ð1; 143Þ ¼ 29:39; MSe ¼ :007636; p <
:001; es ¼ :171; Experiment 1B: F ð1; 71Þ ¼ 15:43;
MSe ¼ :007941; p < :001; es ¼ :178).
PR
OO
F
589 gala2, galb2, guia1, guib1. The test is at http://www.
590 wiu.edu/users/mfbpa/memoryquesab.html
(substitute
591 ‘‘memoryquesabw’’ for the warning condition).
Fig. 1. Mean proportion false recall on suggested versus control items in Experiments 1A and 1B as a function of whether
the narrative in which suggestions were imbedded described the
same event as the witnessed event or a different event. Error
bars are 95% confidence intervals calculated using the MSE
from the Item Type Narrative Type ANOVA (see Loftus &
Masson, 1994).
7
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
CT
CO
R
704 Experiment 2
Experiment 2 was designed with two objectives in
mind. First, we aimed to replicate the central findings of
Experiments 1A and 1B (i.e., a reliable suggestibility
effect both when the narrative described the witnessed
event and when it described another event, but with that
effect larger in the former condition) with a new set of
materials. The event in Experiment 2 was a museum
burglary, and it was presented as a video (rather than a
series of pictures) and the narrative was presented auditorially (rather than as text). Second, we tested the
hypothesis that intrusion rates would be higher when the
event and narrative were thematically similar than when
they were thematically dissimilar. Specifically, we compared performance in three conditions: One in which the
narrative summarized the witnessed event itself (a burglary in a museum), another in which the narrative described a thematically similar but clearly different event
(a burglary in a palace), and a third in which the narrative described a thematically dissimilar event (a school
field trip to a palace).
UN
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
Method
725
Participants
Participants were 185 undergraduate students (63%
women) at the University of Victoria who volunteered
for optional extra credit in an introductory psychology
course. Participants were randomly assigned to six cells:
same-event A (n ¼ 30), same-event B (n ¼ 31), similarevent A (n ¼ 32), similar-event B (n ¼ 32), dissimilarevent A (n ¼ 30), and dissimilar-event B (n ¼ 30). Age
data were not collected.
726
727
728
729
730
731
732
733
734
Materials
An 8-min portion of the movie Return of the Pink
Panther was used as the witnessed (video) event. The
video contained a scene in which a man went through
several different settings in a museum to perform a
burglary. The original sound of the video was replaced
by an accompanying audio narrative describing the
events in the video, which was recorded by the same
womanÕs voice as in the postevent narrative. The sameevent narrative described the museum burglary shown in
the witnessed event video. The similar-event narrative
described a burglary that happened in a palace, whereas
the dissimilar-event narrative described a school field
trip to a palace. There were two versions of each of these
narratives, which differed only with regard to which of
two sets of four suggestions they included (Version A:
‘‘no smoking’’ sign, a mounted deer head, sculpted cupids, a fire escape ladder; Version B: a miniature
building, a sculpted knight, fresco of Greek goddesses,
cans of paint).The suggested details were selected to be
plausible and imagable yet unlikely to be generated as
guesses. As in Experiments 1A and 1B, the suggestions
were of new details not actually witnessed in the event
(rather than subtly different descriptions of witnessed
objects).
An imagery-rating questionnaire asked participants
to assess the vividness of their imagery while listening to
the narrative. The imagery-rating questionnaire was part
of a cover task used to encourage participants to
imagine the people, objects, and actions described in the
narrative, and its results will not be analyzed here.
The memory test questionnaire contained 12 cuedrecall questions: 4 questions concerned details not
present in the video but mentioned in the narrative
(suggested items), 4 concerned details neither present in
the video nor mentioned in the narrative (control items),
and 4 concerned details obviously present in the video
(easy items). In addition to a general instruction to
‘‘Answer the questions based only on what you saw in
the video,’’ each question reminded participants that
they were being asked about the video, not the narrative.
An example critical question is: ‘‘What was sculpted
onto the stand in the big room in the video?’’ (in fact, no
such sculpting appeared in the stand in the video, but the
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
ED
effect of the warning suggests that participants in the
same-event condition did not knowingly rely on the
narrative as a source of answers (i.e., if they had, those
warned that the narrative included inaccuracies should
have less often relied on it).
We propose that memories of suggestions are more
likely to come to mind and be accepted as memories of
the event if the narrative is conceptually similar to the
to-be-remembered event. The extreme of conceptual
similarity is when the narrative is apprehended as a
faithful account of the event itself, but we speculate that
less extreme degrees of conceptual similarity can also
affect the likelihood of source errors. If so, then such a
conceptual source-similarity effect may also be observed
when two different-event narratives are compared: One
that is thematically similar to the visual event and one
that is thematically dissimilar. Even though the surface
form of the specific suggestions would be identical in
these two sorts of narratives, the encoding of those
suggestions should produce memories that are more
likely to come to mind and be accepted at test if the
suggestions were encountered in the context of a similarevent narrative than if they were encountered in the
context of a dissimilar-event narrative. Because both
narratives in this comparison describe a different event
from the witnessed event, there would be equally little
reason for participants to knowingly rely on narrativeonly memories in each condition. Thus higher rates of
intrusions from a similar-event narrative than from a
dissimilar-event narrative would provide additional evidence for the SM framework. This issue was explored in
Experiments 2–4.
RE
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
PR
OO
F
8
No. of pages: 16
DTD 4.3.1 / SPS
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
No. of pages: 16
DTD 4.3.1 / SPS
narrative heard by some participants mentioned that
there were cupids sculpted into the base of the stand).
A 10-min unrelated distracter task was presented
during the delay period between the post-event narrative
and the memory test. In this task, participants were
asked to find as many differences as they could between
two similar pictures.
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
Procedure
Participants were tested individually or in groups of
two to four. After giving informed consent, participants
viewed the museum burglary video in preparation for a
memory test. The video was presented on a television
with the accompanying narrative presented via a cassette
tape. Participants listened to the postevent narrative
immediately after viewing the video, and were asked to
visualize the people, actions, and objects described in the
narrative as vividly as possible. Next, participants rated
their images and then completed the 10-min find-thedifference distracter task.
After the distracter task, participants were given the
memory test. They were asked to base their responses
solely on the video, and to answer every test question
(although 31% left unanswered one or more questions
for which they had received suggestions). Following the
memory test, all participants were fully debriefed.
CT
ED
779
780
781
782
783
784
785
9
PR
OO
F
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
804 Results
Accuracy on filler questions
A between-subjects ANOVA revealed a reliable
effect of narrative type on performance on the four easy
questions,
F ð2; 182Þ ¼ 8:065; MSe ¼ :021; p < :001;
es ¼ :081. Post hoc tests with Bonferroni adjustment
showed that participants in the same-event condition
(M ¼ :98) outperformed participants in the similar-event
condition (M ¼ :88), p < :01, and the dissimilar-event
condition (M ¼ :90), p < :01, with the latter two
conditions not differing, p > :05. The answers to the
easy questions were not mentioned in the narrative, but
it may be that spontaneous recall of those details
while listening to the narrative was more common
with the same-event narrative than with the other
narratives.
820
821
822
823
824
825
826
827
828
829
830
Probability of false recall
Counterbalancing version (A vs. B) had no reliable
main effect in any analysis, nor did it enter into any
reliable interactions (all F s < 1:65). Therefore, all subsequent analyses were collapsed across this variable.
The mean probabilities of false recall are depicted in
Fig. 2. These data were analyzed in a ANOVA with item
type (misinformation reported for the critical and control questions) as a within-subjects variable and narrative type (same, similar, and dissimilar) as a betweensubjects variable. There were significant main effects
UN
CO
R
RE
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
Fig. 2. Mean proportion false recall on suggested versus control items in Experiment 2 as a function of whether the narrative in which suggestions were imbedded described the
witnessed event, a different but thematically similar event, or a
different and dissimilar event. Error bars are 95% confidence
intervals calculated using the MSE from the Item
Type Narrative Type ANOVA (Loftus & Masson, 1994).
of item type, F ð1; 182Þ ¼ 180:5; MSe ¼ :032; p < :01;
es ¼ :498, and narrative type, F ð1; 182Þ ¼ 20:24; MSe ¼
:030; p < :01; es ¼ :182. The interaction between item
type and narrative type was also significant,
F ð2; 182Þ ¼ 21:55; p < :01; es ¼ :191. Because participants almost never spontaneously reported suggested
details as guesses on control items (overall M ¼ :0095),
the subsequent simple effects tests were conducted only
for the items for which participants had received suggestions. The probability of false recall in the same-event
condition exceeded that in the similar-event condition,
(t ¼ 6:01; SE ¼ :044; p < :01) and that in the dissimilar-event condition (t ¼ 4:99; SE ¼ :047; p < :01).
However, contrary to our hypothesis, no reliable difference was found in the rate of false recall between the
similar- and dissimilar-event conditions, t ¼ :71.
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
Discussion
847
Results in the same-event and similar-event narrative
conditions mirrored those of Experiments 1A and 1B: a
reliable misinformation effect was obtained in both
conditions, and that effect was substantially larger in the
same-event narrative condition than in the similar-event
848
849
850
851
852
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
Experiment 3
909
The purpose of Experiment 3 was to compare intrusion rates from similar- and dissimilar-event narratives under conditions that disguised the relationship
between the witnessed event and the narrative. To that
end, the narrative and video were presented to participants in the context of two different experiments, separated by 24 h and conducted in different rooms and by
different experimenters. We presented the narrative in
Session 1 and the video and memory test in Session 2 (cf.
Abeles & Morton, 1999; Holliday & Hayes, 2002;
Lindsay & Johnson, 1989b, for prior examples of such
‘‘reversed’’ misinformation effect procedures). We used
this temporal order for three reasons: First, presenting
the narrative immediately before the test (as in Experiments 1 and 2) likely reduces effects of thematic similarity (because regardless of similarity the recency of the
suggestions would make them highly accessible); a substantial delay between the narrative and test creates
opportunities for similarity (rather than recency) to influence remembering. Second, the delay between narrative and test would likely reduce the likelihood that
subjects spontaneously identify the source of memories
of the narrative when they come to mind (simply because source monitoring declines with delay). Third, the
reversed procedure eliminates the possibility that, while
listening to the narrative, participants would spontaneously note differences between the narrative and the
witnessed event (which might serve to reduce misinformation effects; cf. Tousignant, Hall, & Loftus, 1986).
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
Method
939
Design
In Session 1, participants listened to an audio narrative. In Session 2 (conducted 24 h later, in a different
room and with a different experimenter) participants
viewed the video and took the memory test. The narrative presented in Session 1 was either thematically similar
or thematically dissimilar to the witnessed event video.
Two versions of each type of narrative were created,
differing in terms of which of two sets of six suggestions
they contained. Thus there were four cells in the design:
similar-event A (n ¼ 16), similar-event B (n ¼ 16), dissimilar-event A (n ¼ 16), dissimilar-event B (n ¼ 19).
Unlike Experiments 1A, 1B, and 2, there were no true
control questions on the memory test. Rather, each
participant was exposed to one of two suggestions regarding each of six critical items. The guessing baseline
was the number of questions on which each participant
guessed suggested details from the non-presented narrative divided by the number of questions on which that
participant did not report suggested details from the
presented narrative. This method enabled us to increase
the number of suggestions given to each participant;
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
CO
R
RE
CT
ED
narrative condition. Contrary to expectations, however,
the intrusion rate in the similar-event narrative condition was no higher than that in the dissimilar-event
narrative condition. On the face of it, this null effect
joins those reported by Bonto and Payne (1991), Shaw et
al. (1997), and Mitchell and Zaragoza (2001) as evidence
against the SM frameworkÕs prediction that, all else
being equal, false reports should increase with source
similarity. We conducted several versions of Experiment
2 (varying the materials and instructions across experiments), and consistently failed to find a reliable difference between similar- and dissimilar-event narrative
conditions.
We had hypothesized that memories of suggestions
encountered in a similar-event narrative should be
processed in ways that make them more similar to
memories of the event itself, and hence more likely to
come to mind and be accepted as answers at test. Why
was this not the case? Perhaps because these effects were
attenuated by situational factors that led participants to
link the event and narrative even in the dissimilar-narrative condition. In Experiment 2, the event, narrative,
and test were all presented close together in time and as
parts of a single experiment, conducted in the same
place, and presented by the same experimenter. Also,
despite having a different theme, the dissimilar-events
narrative had considerable overlap with the visual event
(e.g., both could be described as a tour through a fancy
building filled with numerous art objects) and with the
test questions. The concomitance between event and
narrative, coupled with the overlapping semantic content between event and narrative and between test and
narrative, likely made the relationship between the two
sources patently obvious even to participants in the
dissimilar events condition. In turn, the obviousness of
the relationship between the two sources may have
overwhelmed any effect of our manipulation of the
thematic similarity between the narrative and the event
on intrusion rates. That is, when event, narrative, and
test are concomitant, participants are likely to apprehend the two sources as related even in the dissimilarevents condition, and hence there will be no effect of
thematic similarity. Spencer and Weisberg (1986) reported a related effect on analogical transfer: When
there was a delay between the training and target
problems, transfer was much greater when the two
phases were presented as parts of a single experiment
than when the target problem was presented as an ostensibly unrelated lecture demonstration; when there
was no delay, in contrast, transfer was equivalently high
regardless of the context manipulation. If concomitance
likewise trumps thematic similarity in our suggestibility
procedure, then separating the narrative and the event
so as to disguise the relationship between the two
should yield an effect of thematic similarity on intrusion
rates.
UN
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
PR
OO
F
10
No. of pages: 16
DTD 4.3.1 / SPS
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
No. of pages: 16
DTD 4.3.1 / SPS
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
Participants
Participants were 67 undergraduate students (69%
women) at the University of Victoria, who volunteered
for optional extra credit in an introductory psychology
course or for $10 ($5 for each session). Age data were
not collected. Participants were randomly assigned to
the four cells of the design and were tested individually
or in groups of two to three.
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
Materials
As in Experiment 2, the similar-event narrative described a palace burglary, whereas the dissimilar-event
narrative described a school field-trip to a palace. Each
narrative was approximately 1200 words long, and included six suggestions. Version A and Version B of the
narratives differed only in which of the six suggestions
they contained, and these suggestions were counterbalanced across participants (Version A: ‘‘no smoking’’
sign, painting of a prince, statue of a Greek goddess,
medieval swords, sculpted dolphins, fresco of knights in
armor; Version B: ‘‘no photography’’ sign, painting of a
queen, statue of a man wearing armor, medieval shield,
sculpted cupids, fresco of David and Goliath). The
suggestions were of new objects, not present in the video. The witnessed event video was the same video clip
of a museum burglary used in Experiment 2. In the
current experiment, however, the video clip and accompanying narrative were digitized. An additional filler
video was used as part of the cover task for Session 2.
This filler video was a 9-min excerpt (with the original
sound track) from the IMAX movie Everest, which
showed hikers climbing Mount Everest. None of the
critical details in the witnessed event or the narratives
appeared in the filler video. Both videos were presented
on a 15-in. computer screen with external speakers.
Three questionnaires were used for this experiment:
An auditory perception questionnaire, a visual perception questionnaire, and a memory test questionnaire.
The first two questionnaires were part of the cover tasks
and results from them will not be reported. The auditory
perception questionnaire asked participants to rate their
perceptions of the auditory components of the two
videos. The visual imagery questionnaire asked participants to rate their imagery while listening to the narrative. The memory test questionnaire consisted of nine
cued recall questions like those used in Experiment 2. Six
questions concerned the critical details presented only in
the narrative, while the other three questions were easy
filler questions regarding details obvious in the video.
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
Results
1050
Accuracy on filler questions
Performance on the filler questions was near ceiling
and did not reliably differ between the similar-event
condition (M ¼ :96) and the dissimilar-event condition
ðM ¼ :92Þ; t ¼ 1:1.
1051
1052
1053
1054
1055
Probability of false recall
In the current experiment, participants who listened
to narrative version A served as the control group for
participants who listened to narrative version B, and
vice versa. Therefore, if a participant who heard narrative version A spontaneously reported a critical detail
that was mentioned only in narrative version B, it was
counted as false recall in the control condition (and indexed as the proportion of questions on which suggested
details were not reported that such guesses were reported). Counterbalancing version (A vs. B) had no reliable main effect in any analysis nor did it enter into any
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
UN
CO
R
RE
CT
ED
964
965
966
967
968
969
970
971
room on the third-floor of the psychology building.
After giving informed consent, participants listened to
the misinformation narrative, with instructions to visualize what was described in the narrative. Afterwards,
participants rated how vivid their images were. Participants then listened to the narrative again, with instructions to visualize it in vivid colors, and then rated how
colorful their images were. The experimenter then provided a bogus debriefing, according to which the experiment was designed to determine whether people
imagine events in color spontaneously and whether they
can imagine in color when instructed to do so. The true
purpose of the imagery instructions and the repeated
presentation of the narrative was to enhance participantsÕ memory for the details in the narrative. All participants were awarded credit or $5 at the conclusion of
Session 1.
Session 2 was conducted by a man in a room in the
basement of the psychology building. Participants were
told that they would view a video clip depicting a burglary, with the original soundtrack removed and replaced by a womanÕs voice narrating the action,
followed by a second video clip about mountaineering
with its original soundtrack. After participants viewed
the critical video, they rated the loudness of the event
depicted in the video. Participants then viewed the filler
video and answered the same question about that video
(as part of the cover task). Finally, participants were
given the surprise memory test. Participants were told to
base their responses solely on what they saw on the video, and to answer every question (although 28.3% left
one or more questions unanswered). Following the
completion of the memory test, participants were thoroughly debriefed and awarded their credit or $5 for
Session 2.
PR
OO
F
962 note that Experiment 2 established that the guessing
963 baseline for these items was near zero.
11
1012 Procedure
1013
The experiment consisted of two sessions scheduled
1014 24 h apart. Session 1 was conducted by a woman in a
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
items in both the similar-event condition (t ¼ 7:19;
SE ¼ :042) and the dissimilar-event condition (t ¼ 5:2;
SE ¼ :029).
1089
1090
1091
Discussion
1092
As predicted, intrusion rates were substantially
higher in the similar-event narrative condition than in
the dissimilar-event narrative condition. Further discussion of this finding is postponed pending report of
Experiment 4.
1093
1094
1095
1096
1097
Experiment 4
1098
In Experiment 2 (in which all parts of the procedure
were conducted in a single session), no effect of thematic
similarity was obtained. In contrast, in Experiment 3 (in
which the narrative and witnessed event were presented
in two separate contexts) false reports of suggestions
were substantially more common when the witnessed
event and narrative were thematically similar than when
they were thematically dissimilar. We take this as evidence that disguising the relationship between the two
sources creates an opportunity for effects of thematic
source-similarity to arise. Of course, this is an acrossexperiment comparison, and Experiments 2 and 3 differed in several ways in addition to what we take to be
the key difference. Experiment 4 was therefore designed
as a conceptual replication of Experiments 2 and 3. Half
of the participants in Experiment 4 were tested in a
procedure that, like Experiment 2, would likely make the
relationship between the event and narrative obvious,
whereas the remaining participants were tested in a
procedure that, like Experiment 3, disguised the relationship between the two sources. We predicted no effect
of thematic similarity in the 1-Day (i.e., non-disguised)
conditions, but more intrusions from the similar-events
narrative than from the dissimilar-events narrative in the
2-Day (disguised) conditions.
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
Method
1124
Participants
Ninety-six University of Victoria undergraduate
students (63% women; age range from 17 to 40 years,
mean age ¼ 19.89, SD ¼ 3:65) participated in return for
optional extra-credit points in an introductory psychology course. There were 24 participants in each condition: 1-Day similar, 1-Day dissimilar, 2-Day similar, and
2-Day dissimilar. Participants were randomly assigned
to condition and tested individually.
1125
1126
1127
1128
1129
1130
1131
1132
1133
Materials and procedure
The materials in this experiment were the same as in
Experiment 3. Participants were tested individually, half
1134
1135
1136
ED
reliable interactions (all F s < 1:61). Therefore, all subsequent analyses were collapsed across this variable.
The mean rates of false recall are depicted in Fig. 3.
These data were analyzed in a ANOVA with narrative
type (similar vs. dissimilar) as a between-subjects variable and item type (critical vs. control) as a withinsubjects variable. Participants more often reported a
critical misleading detail when they had heard it in the
narrative than when they had not, F ð1; 65Þ ¼ 80:94;
MSe ¼ :021; p < :01; es ¼ :56. Those in the similarevent condition recalled more misinformation than did
those in the dissimilar-event condition, F ð1; 65Þ ¼ 7:83;
MSe ¼ :026; p < :01; es ¼ :108. The interaction between narrative type and item type was also significant,
F ð1; 65Þ ¼ 8:58; p < :01; es ¼ :12. FisherÕs protected
simple effects t-tests of the rates of false recall of suggested details showed that, as predicted, the similarevent narrative generated significantly more false recall
than the dissimilar-event narrative, t ¼ 3:0; SE ¼ :050.
The rate of false recall of suggested items was significantly greater than the rate of false guesses of control
UN
CO
R
RE
CT
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
PR
OO
F
12
No. of pages: 16
DTD 4.3.1 / SPS
Fig. 3. Mean proportion false recall in Experiment 3 as a
function of whether the narrative in which suggestions were
imbedded described an event that was thematically similar or
dissimilar to the witnessed event. Error bars are 95% confidence
intervals calculated using the MSE from the Item
Type Narrative Type ANOVA (see Loftus & Masson, 1994).
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
No. of pages: 16
DTD 4.3.1 / SPS
Fig. 4. Mean proportion false recall in the 1-Day and 2-Day
conditions of Experiment 4 as a function of whether the narrative in which suggestions were imbedded described an event
that was thematically similar or dissimilar to the witnessed
event. Error bars are 95% confidence intervals calculated using
the MSE from the Item Type Narrative Type ANOVA (see
Loftus & Masson, 1994).
ED
in a single session and half in two sessions. For participants in the 2-Day condition, the procedure replicated
Experiment 3 (except that both experimenters were
women). Session 1 was conducted in a basement room
and participants were told that they were participating
in an imagery study. They were informed that they
would hear a narrative twice and then be asked to
complete a questionnaire about their imagery for that
narrative. After completing these tasks, participants received a bogus debriefing. The following day, participants arrived at a third-floor testing room to take part in
an ostensibly unrelated study. The experimenter (a different person from on day one) explained that the study
examined how audio soundtracks influence visual perception. Participants watched the critical video and filler
video and completed the filler questionnaire and surprise
memory test as in Experiment 3. The cued-recall questions were the same as in Experiment 3, and as in the
prior experiments participants were asked to base their
responses solely on what they had seen in the video and
to answer every question (although 27% of participants
left one or more questions unanswered). Upon completion of the experiment, participants were debriefed. The
same procedure was used in the 1-Day condition, except
that one experimenter conducted both sessions in the
basement room, with the critical and filler videos shown
immediately following the narrative.
CT
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
13
PR
OO
F
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
1164 Results
Accuracy on filler questions
Similar to Experiments 2 and 3, participants were
highly accurate on the three filler questions (mean accuracy ¼ 94%). There were no significant effects of delay
or narrative similarity on accuracy (pÕs > .254), nor was
there a reliable interaction (p ¼ :467).
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
Probability of false recall
A preliminary ANOVA included counterbalancing
version (A vs. B) as an independent variable, along with
item type, similarity, and day. There was a non-significant tendency toward a main effect of counterbalancing
narrative, F ð1; 88Þ ¼ 3:48; MSe ¼ :043; p ¼ :066, but
that factor did not enter into any remotely reliable interactions (all pÕs > .34) so the analyses below collapse
across the counterbalancing factor. Fig. 4 depicts the
data.
The proportion of items on which participants reported suggestions, and the proportion of other items on
which they guessed non-presented (control) suggestions,
were analyzed in a 2 (item type) 2 (similarity) 2 (1Day vs. 2-Day) ANOVA in which the first of these
variables was a repeated measure. The key finding was
the predicted three-way interaction between these
variables, F ð1; 92Þ ¼ 4:10; p ¼ :046; MSe ¼ :041; es ¼
:043. Separate Item Type Similarity ANOVAs for the
UN
CO
R
RE
1165
1166
1167
1168
1169
1170
1- and 2-Day conditions showed that there was a reliable
effect of item type in each condition (1-Day,
F ð1; 46Þ ¼ 22:94; p ¼ :000; MSe ¼ :035; es ¼ :33; 2-Day,
F ð1; 46Þ ¼ 39:18;
p ¼ :000; MSe ¼ :046; es ¼ :46).
Moreover, just as predicted, in the 1-Day condition
there was no Item type Similarity interaction (F < 1),
whereas in the 2-Day condition that interaction was
reliable, F ð1; 46Þ ¼ 8:67; p ¼ :005; MSe ¼ :046; es ¼ :16.
1190
1191
1192
1193
1194
1195
1196
1197
Discussion
1198
As predicted, the 1-Day condition replicated the null
effect of thematic similarity obtained in Experiment 2,
whereas the 2-Day condition replication the reliable effect of thematic similarity obtained in Experiment 3.
Evidently, when event, narrative, and test are all presented as part of the same overall experience, the high
degree of concomitance overwhelms any effect of thematic similarity, presumably because participants apprehend the narrative as being related to the event even
in the dissimilar-events condition. In contrast, when the
relationship between event and narrative is disguised, a
thematic-similarity effect emerges.
Why did the interaction take the form of a particularly high rate of false reports in the 2-Day/similarevents condition, rather than a particularly low rate of
false reports in the 2-Day/dissimilar-events condition?
We speculate that in the 1-Day condition suggestions
very often came to mind at test, due to the semantic
overlap between questions and suggestions, the recency
of suggestions, and the fact that the situation led participants to apprehend the event and narrative as related. It is likely, though, that participants in the 1-Day
conditions often correctly attributed those memories to
the narrative, and hence did not rely on them as answers
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
to the questions about the event. Together, these two
phenomena led to a relatively low rate of false-reports of
suggestions in the 1-Day conditions that did not vary as
a function of thematic similarity. In the 2-Day condition, in contrast, thematic similarity played a role in
increasing both the likelihood that suggestions would
come to mind at test and that they would be accepted as
answers, and the 24-h delay (and other aspects of the
disguise manipulation) likely lowered the likelihood that
participants would spontaneously attribute those memories to the narrative. Consequently, the similar-events
narrative produced a high rate of intrusions in the 2-Day
condition.
1236 General discussion
CO
R
RE
CT
ED
The five experiments reported above replicated the
core finding of Allen and Lindsay (1998): details mentioned in a narrative description of one event sometimes
intruded into participantsÕ reports of a different visual
event. These findings substantially increase the applied
scope of the eyewitness misinformation effect. Whereas
only some witnesses are exposed to misinformation in
the context of statements about a witnessed event, virtually all witnesses have past experiences (direct or vicarious) that include details that map onto questions
that might be asked in a forensic context. It is worth
noting that the false reports in these studies did not
merely involve subtle distortions of witnessed details
(e.g., misremembering Coke as Pepsi) but rather reports
of entire objects that were not witnessed at all.
Consistent with the source-monitoring framework,
Experiments 1A, 1B, and 2 found that intrusions were
more frequent when the narrative was about the witnessed event than when it was about another event.
Presumably, apprehending the narrative as a description
of the witnessed event leads participants to think about
it during encoding in ways that are similar to how they
thought about the witnessed event. In consequence of
that high similarity, suggestions are both likely to come
to mind and likely to be accepted as memories of the
event during test. The null effect of the warning in Experiments 1A and 1B suggests that this is not merely a
matter of participants more often knowingly relying on
narrative-only memories in the same-event condition.
In contrast to the robust difference in the rate of intrusions from same-event versus different-event narratives, Experiment 2 found no reliable difference in the
rate of intrusions from narratives of thematically similar
versus dissimilar events. Importantly, however, the intrusion rate was substantially higher in the similar- than
dissimilar-events condition of Experiment 3, in which
the relationship between the narrative and event was
disguised by presenting the two sources in the context of
ostensibly separate experiments. Experiment 4 replicated
UN
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
that thematic-similarity effect in the 2-Day condition,
and also replicated the null effect of Experiment 2 in the
1-Day condition.
These results, together with findings from prior
studies, suggest that whether source-similarity manipulations influence suggestibility effects depends on several
factors. First, test instructions must be such that participants are unlikely to report information they know
came only from the extra-event source: If participants
deliberately use the extra-event source as a basis for
answers, source-similarity is unlikely to affect intrusion
rates. Second, the test must allow participants to report
suggested details (as opposed, e.g., to the McCloskey &
Zaragoza, 1985, modified test). Third, the source-similarity manipulation must be sufficiently powerful in the
context of other aspects of the task. Some manipulations
(e.g., Zaragoza & LaneÕs, 1994 manipulations of the
extent to which participants elaborated and integrated
suggestions, and our manipulation of same- vs. differentevent narratives) are so powerful that effects can be
obtained even when all phases of the experiment are
conducted in a single session. Other manipulations (e.g.,
our manipulation of thematic similarity) may not produce detectable effects when numerous other aspects of
the situation highlight the relationship between the event
and extra-event information (as in our 1-Day conditions), but yield effects when these other dimensions of
background similarity are attenuated to disguise the
relationship between event and narrative (as in our 2Day conditions).
The variation across our experiments, and across
previous experiments, in whether source-similarity manipulations affected suggestibility are consistent with the
idea that multiple memorial and situational dimensions
interact to determine source-monitoring performance.
Event memories are complex and multifaceted, and the
aspects of memories that are revived and contribute to
source attributions vary across situations. Consequently,
it is often difficult to predict the effect of manipulating a
single dimension of source-relevant information. For
example, a conceptual replication of Shaw et al.Õs (1997)
procedure, with the manipulation of same versus different language during the target event and extra-event
information conducted across two ostensibly unrelated
sessions to disguise their relationship, might yield a
source-similarity effect (i.e., more false reports in the
same-languages condition than in the different-languages condition). But it might be that no such effect
would be obtained, because people are very skilled at
abstracting meanings from the surface form of the language in which they are communicated, and direct tests
of memory tend to draw on memory for relatively abstract, conceptual operations (Toth, 2000).
The development of better-specified models of source
monitoring depends, in part, on progress toward understanding the crucial but elusive psychological con-
PR
OO
F
14
No. of pages: 16
DTD 4.3.1 / SPS
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
No. of pages: 16
DTD 4.3.1 / SPS
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
PR
OO
F
mental Psychology: Human Experimental Psychology, 52a,
581–592.
Abra, J. C. (1972). List differentiation and forgetting. In C. P.
Duncan, L. Sechrest, & A. W. Melton (Eds.), Human
memory: Festschrift for Benton J. Underwood (pp. 25–57).
New York: Appleton-Century-Crofts.
Allen, B. P. (1999). Shadows as sources of cues for distance of
shadow casting objects. Perceptual and Motor Skills, 89,
571–584.
Allen, B. P., & Lindsay, D. S. (1998). Amalgamations of
memories: Intrusion of information from one event into
reports of another. Applied Cognitive Psychology, 12, 277–
285.
Birnbaum, M. H. (Ed.). (2000). Psychological experiments on
the Internet. San Diego: Academic Press.
Bonto, M. A., & Payne, D. G. (1991). Role of environmental
context in eyewitness memory. American Journal of Psychology, 104, 117–134.
Chambers, K. L., & Zaragoza, M. S. (2002). Intended and
unintended effects of explicit warnings on eyewitness
suggestibility: Evidence from source identification tests.
Memory & Cognition, 29, 1120–1129.
Chandler, C. C., Gargano, G. J., & Holt, B. C. (2001).
Witnessing postevents does not change memory traces, but
can affect their retrieval. Applied Cognitive Psychology, 15,
3–22.
Eberman, C., & McKelvie, S. J. (2002). Vividness of visual
imagery and source memory for audio and text. Applied
Cognitive Psychology, 16, 87–95.
Frost, P., Ingraham, M., & Wilson, B. (2002). Why misinformation is more likely to be recognized over time: A source
monitoring account. Memory, 10, 179–185.
Hekkanen, S. T., & McEvoy, C. (2002). False memories and
source-monitoring problems: Criterion differences. Applied
Cognitive Psychology, 16, 73–85.
Hicks, J. L., & Marsh, R. L. (2001). False recognition occurs
more frequently during source identification than during
old-new recognition. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 27, 375.
Holliday, R. E., & Hayes, B. K. (2000). Dissociating automatic
and intentional processes in childrenÕs eyewitness memory.
Journal of Experimental Child Psychology, 75, 1–42.
Holliday, R. E., & Hayes, B. K. (2001). Automatic and
intentional processes in childrenÕs eyewitness suggestibility.
Cognitive Development, 16, 617–636.
Holliday, R. E., & Hayes, B. K. (2002). Automatic and
international processes in childrenÕs recognition memory:
The reversed misinformation effect. Applied Cognitive Psychology, 16, 1–16.
Jacoby, L. L., Kelley, C. M., & Dywan, J. (1989a). Memory
attributions. In H. L. Roediger & F. I. M. Craik (Eds.),
Varieties of memory and consciousness: Essays in honor of
Endel Tulving (pp. 391–422). Hillsdale, NJ: Erlbaum.
Jacoby, L. L., Woloshyn, V., & Kelley, C. M. (1989b).
Becoming famous without being recognized: Unconscious
influences of memory produced by dividing attention.
Journal of Experimental Psychology: General, 118, 115–125.
Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993).
Source monitoring. Psychological Bulletin, 114, 3–28.
Johnson, M. K., & Raye, C. L. (2000). Cognitive and brain
mechanisms of false memories and beliefs. In D. L. Schacter
CO
R
RE
CT
ED
struct of similarity. Even in quite simple situations (e.g.,
explicit similarity ratings of pairs of pictures or words),
perceptions of similarity are complex. One reason for
this is that even relatively simple objects, such as twodimensional geometric shapes, have an indefinitely large
number of features or aspects (e.g., angularity, complexity, size, color, and location) and the relative
weightings of the contributions of different dimensions
to perceptions of similarity varies across contexts, instructions, and individuals (e.g., Medin, Goldstone, &
Gentner, 1993). More fundamentally, similarity is not
necessarily based on physical features, but can be based
on more abstract characteristics such as functions (e.g.,
hammer and rock) or thematic relations (milk and cow)
(e.g., Sloman & Rips, 1998; Wisniewski & Bassok,
1999). As Medin et al. (1993) argued, similarity is not a
property of objects, but rather a product of mental
processes. Little is known of how mental processes determine degrees of similarity between complex, temporally extended, naturalistic experiences (e.g., witnessing
a crime and hearing a story).
Predicting source-monitoring performance is further
complicated by the fact that source-monitoring accuracy
is not determined solely by the degree of similarity between memories from different sources. Rather, source
monitoring is also influenced by the remembererÕs goals
and orientation at test, decision-making strategies and
biases, beliefs and expectations, accessibility, etc. (see
Lindsay & Johnson, 2001, pp. 151–152). Thus a complete understanding of source monitoring requires
specification of how multiple aspects of memories and
situations interact to determine SM accuracy.
The research reported here does not solve these
problems, but it does take modest steps in that direction
by showing that the effect of a similarity manipulation
on intrusion rates depends on contextual variables (i.e.,
whether the context does or does not lead participants to
apprehend two sources as related). Exploring how various aspects and dimensions of past experiences interact
with characteristics of the memory testing situation to
modulate source-attribution accuracy is a long-term
goal for future research (see also Lindsay & Read, 2001).
For now, we can conclude that (a) people sometimes
intrude details from a narrative of one event into their
reports of a different witnessed event, although (b) they
do so less often than when the narrative describes the
witnessed event itself, and (c) if the relationship between
the two different-event sources is disguised at acquisition
then across-event intrusions are more likely when the
two sources share similar themes.
UN
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382 References
1383 Abeles, P., & Morton, J. (1999). Avoiding misinformation:
1384
Reinstating target modality. Quarterly Journal of Experi-
15
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
YJMLA 2948
ARTICLE IN PRESS
DISK / 1/10/03 / Sheela(CE) / Manju (TE)
McGraw, K. O., Tew, M. D., & Williams, J. E. (2000). The
integrity of Web-delivered experiments: Can you trust the
data? Psychological Science, 11, 502–506.
Medin, D. L., Goldstone, R. L., & Gentner, D. (1993). Respects
for similarity. Psychological Review, 100, 254–278.
Mitchell, K. J., & Zaragoza, M. S. (1996). Repeated exposure to
suggestion and false memory: The role of contextual
variability. Journal of Memory and Language, 35, 246–
260.
Mitchell, K. J., & Zaragoza, M. S. (2001). Contextual overlap
and eyewitness suggestibility. Memory & Cognition, 29, 616–
626.
Postman, L., & Stark, K. (1969). Role of response availability
in transfer and interference. Journal of Experimental Psychology, 79, 168–177.
Roediger III, H. L., Jacoby, J. D., & McDermott, K. B. (1996).
Misinformation effects in recall: Creating false memories
through repeated retrieval. Journal of Memory and Language, 35, 300–318.
Shaw III, J. S., Garcia, L. A., & Robles, B. E. (1997). Crosslanguage postevent misinformation effects in Spanish–
English bilingual witnesses. Journal of Applied Psychology,
82, 889–899.
Sloman, S. A., & Rips, L. J. (1998). Similarity as an explanatory
construct. Cognition, 65, 87–101.
Spencer, R. M., & Weisberg, R. W. (1986). Context-dependent
effects on analogical transfer. Memory & Cognition, 14, 442–
449.
Toth, J. P. (2000). Nonconscious forms of human memory. In
E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of
memory (pp. 245–261). London: Oxford University
Press.
Tousignant, J. P., Hall, D., & Loftus, E. F. (1986). Discrepancy
detection and vulnerability to misleading postevent information. Memory & Cognition, 14, 329–338.
Vecera, S. P., Vogel, E. K., & Woodman, G. F. (2002). Lower
region: A new cue for figure-ground assignment. Journal of
Experimental Psychology: General, 131, 194.
Wisniewski, E. J., & Bassok, M. (1999). What makes a
man similar to a tie? Stimulus compatibility with comparison and integration. Cognitive Psychology, 39, 208–
238.
Whittlesea, B. W. A. (2003). On the construction of behavior
and subjective experience: The production and evaluation of
performance. In J. S. Bowers & C. J. Marsolek (Eds.),
Rethinking implicit memory (pp. 239–260). London: Oxford
University Press.
Zaragoza, M. S., & Koshmider, J. W. (1989). Misled subjects
may know more than their performance implies. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 15, 246–255.
Zaragoza, M. S., & Lane, S. M. (1994). Source misattributions
and the suggestibility of eyewitness memory. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 20, 934–945.
CO
R
RE
CT
ED
& E. Scarry (Eds.), Memory, brain, and belief (pp. 35–86).
Cambridge, MA: Harvard University Press.
Keppel, G., & Underwood, B. J. (1962). Retroactive inhibition
of R–S associations. Journal of Experimental Psychology,
64, 400–404.
Lindsay, D. S. (1990). Misleading suggestions can impair
eyewitnessesÕ ability to remember event details. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 16, 1077–1083.
Lindsay, D. S., Gonzales, V., & Eso, K. (1995). Aware and
unaware uses of memories of postevent suggestions. In M. S.
Zaragoza, J. R. Graham, G. C. N. Hall, J. E. Hirschman, & Y.
S. Ben-Porath (Eds.), Memory and testimony in the child
witness (pp. 86–108). Thousand Oaks, CA: Sage Publications.
Lindsay, D. S., & Johnson, M. K. (1989a). The eyewitness
suggestibility effect and memory for source. Memory &
Cognition, 17, 349–358.
Lindsay, D. S., & Johnson, M. K. (1989b). The reversed
eyewitness suggestibility effect. Bulletin of the Psychonomic
Society, 27, 111–113.
Lindsay, D. S., & Johnson, M. K. (2001). False memories, fuzzy
trace theory, and the source monitoring framework. Learning and Individual Differences, 12, 145–161.
Lindsay, D. S., Johnson, M. K., & Kwon, P. (1991). Developmental changes in memory source monitoring. Journal of
Experimental Child Psychology, 52, 297–318.
Lindsay, D. S., & Read, J. D. (2001). The recovered memories
controversy: Where do we go from here?. In G. Davies & T.
Dalgleish (Eds.), Recovered memories: Seeking the middle
ground (pp. 71–94). London: Wiley.
Loftus, E. F., Miller, D. G., & Burns, H. J. (1978). Semantic
integration of verbal information into a visual memory.
Journal of Experimental Psychology: Human Learning and
Memory, 4, 19–31.
Loftus, G. R., & Masson, M. E. J. (1994). Using confidence
intervals in within-subject designs. Psychonomic Bulletin &
Review, 1, 476–490.
Marsh, R. L., & Hicks, J. L. (1998). Test formats change
source-monitoring decision processes. Journal of Experimental Psychology: Learning, Memory, and Cognition, 24,
1137–1151.
McCloskey, M., & Zaragoza, M. (1985). Misleading postevent
information and memory for events: Arguments and
evidence against memory impairment hypotheses. Journal
of Experimental Psychology: General, 114, 1–16.
McGeoch, J. A. (1936). Retroactive inhibition as a function of
the relative amounts of original and interpolated materials.
Psychological Bulletin, 33, 613–614.
McGeoch, J. A., & McGeoch, G. O. (1937). Studies in
retroactive inhibition: X. The influence of similarity of
meaning between lists of paired associates. Journal of
Experimental Psychology, 21, 320–329.
McGovern, J. B. (1964). Extinction of associations in four
transfer paradigms. Psychological Monographs: General and
Applied, 78, 21.
UN
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
D. Stephen Lindsay et al. / Journal of Memory and Language xxx (2003) xxx–xxx
PR
OO
F
16
No. of pages: 16
DTD 4.3.1 / SPS
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555