Download Recall, Recognition, and Confidence Patterns

APPLIED COGNITIVE PSYCHOLOGY
Appl. Cognit. Psychol. 13: 257±268 (1999)
Recall, Recognition, and Con®dence Patterns
in Eyewitness Testimony
MALEN MIGUELES* and ELVIRA GARCIÂA-BAJOS
University of the Basque Country, San SebastiaÂn, Spain
SUMMARY
The diversity of methods, contents and tests used in the study of eyewitness memory may have
contributed to discrepancies in results in this ®eld. In this experiment, using incidental or
intentional learning, we examine the recall and recognition of actions and details concerning the
central and peripheral information of a kidnapping. A similar pattern emerges in free recall, hits
and recognition con®dence: scores are much higher in actions than in central details and there
are almost no di€erences between peripheral actions and details, showing that the distribution of
attentional resources is di€erent for actions and details in central than in peripheral information.
Although in recall the degree of error was low, in recognition false alarms, especially those
in central actions, reduced the level of accuracy to even lower than chance performance in
both incidental and intentional groups, also showing that subjects accept false but plausible
contents with a high level of con®dence. Copyright # 1999 John Wiley & Sons, Ltd.
Although research on eyewitness testimony has attempted to determine memory
accuracy and completeness in emotional events, the diversity of research methods
used, the lack of uniformity in contents evaluated and the di€erent tests utilized
have hindered the elaboration of a coherent model. Thus, ®eld studies (Cutshall and
Yuille, 1989; Yuille and Cutshall, 1986) and work on autobiographical emotional
or traumatic experiences (Reisberg et al., 1988; Wagenaar and Groeneweg, 1990)
indicate that subjects have vivid outstanding recall of events and good long-term
retention. In contrast, others present recall impairment when there is increased
violence (Kuehn, 1974), amnesia in traumatic situations (Christianson and Nilsson,
1989) and lower performance in emotional as opposed to neutral versions in
laboratory work where slides or ®lms are used to simulate events (Cli€ord and Hollin,
1981; Cli€ord and Scott, 1978; Loftus and Burns, 1982).
Studies in which positive emotional e€ects are noted mainly evaluate central aspects
of events, whereas negative e€ects are re¯ected when assessing more particular details
or, as Christianson and Loftus (1991) point out, when central and peripheral aspects
are combined. It is logical, except in the case of highly traumatic events or in clinical
cases, that emotional contents draw attention to the central aspects of the event,
*Correspondence to: Malen Migueles, Faculty of Psychology, The University of the Basque Country,
Avda Tolosa 70, 20009 San SebastiaÂn, Spain. E-mail: [email protected]
Contract grant sponsor: The University of the Basque Country
CCC 0888±4080/99/030257±12 $17.50
Copyright # 1999 John Wiley & Sons, Ltd.
Accepted 20 March 1998
258
M. Migueles and E. GarcõÂa-Bajos
leaving the more peripheral contents outside of the focus of attention, particularly
if resources are limited. Therefore, the di€erence between central and peripheral
information set forth in the attentional narrowing hypothesis (Easterbrook, 1959) may
largely explain the divergences detected. Taking this variable into account, ®eld
studies (Christianson and HuÈbinette, 1993), traumatic autobiographical experiences
(Christianson and Loftus, 1990; Wessel and Merckelbach, 1994), laboratory work
simulating events by means of slides (Burke et al., 1992; Christianson, 1984; Christianson et al., 1996; Christianson and Loftus, 1987, 1991; Heuer and Reisberg, 1990), ®lms
(Kebeck and Lohaus, 1986) or aversive situations (Wessell and Merckelbach, 1997)
demonstrate that central contents of emotional events are recalled with precision to the
detriment of other more peripheral contents.
The distinction between central and peripheral information is pertinent, although
not without complications. The main ones lies in its de®nition. Christianson
(e.g. Christianson, 1992; Christianson et al., 1996) embraces a perceptual-spatial
de®nition, regarding aspects linked to the emotional source as central (gist and its
central details), and contents preceding or following the emotional event, or those
irrelevant or spatially non-central to the emotional core, as peripheral. However,
Burke, Heuer and Reisberg (1992; Heuer and Reisberg, 1990, 1992) support a
conceptual de®nition, viewing aspects which cannot be excluded or changed without
altering the argument of the event as central, and contents irrelevant to the argument
as peripheral. An additional problem involves determining which contents should be
evaluated so that performance in both types of information (central versus peripheral)
is equal, since gist, facts and relevant details are not comparable to more particular
details. Laboratory studies have placed emphasis on details (Heuer and Reisberg,
1992), but we cannot exclude the recall of actions, de®ned as the clearly apparent
execution of verbal and non-verbal behaviours (Sanders and Chiu, 1988), since they
accurately capture the argument and sequence of facts.
The object of this experiment is to study recall and recognition of actions and
details in both central and peripheral information. A ®lm depicting a kidnapping
attempt was selected; it had a highly emotional content that captivated the viewer's
attention from beginning to end. The kidnapping itself (central information), which
occurs quickly and unexpectedly, appears in the central scenes; but beforehand,
afterwards and simultaneously, there is a sequence of incidents that, although not
directly connected with the emotional source (peripheral information), make up part
of the event.
In laboratory studies an attempt has been made to prevent eyewitnesses being
intentional material witnesses, indicating to them, for example, that the object of the
study is to examine their physiological responses (e.g. Burke et al., 1992; Christianson,
1984; Heuer and Reisberg, 1990). In this study one intentional and one incidental
learning condition were prepared for the purposes of checking to see if the type of
learning a€ects general performance or the memory pattern for the di€erent contents
of the event.
Another relevant aspect in the area of eyewitness memory is the repetition of
evaluation tests, especially in recalling or narrating the event. Repetition improves net
recall (Dunning and Stern, 1992; Scrivner and Safer, 1988; Turtle and Yuille, 1994),
but can also produce errors and intrusions (Eugenio et al., 1982). Therefore, it seems
appropriate to ascertain whether the e€ects are the same for actions and details in
both central and peripheral information.
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
Eyewitness testimony
259
Finally, the study of response con®dence has been restricted almost exclusively to
eyewitness identi®cation. In this case, by means of a true/false recognition test, we can
check to see if con®dence in testimony is determined by the type of information
(central, peripheral) and contents (actions, details) evaluated, or if, on the contrary, it
is determined by the four possible types of responses in recognition: Hits, False
Alarms, Correct Rejections and Misses.
METHOD
Subjects
A total of 102 psychology students from the University of the Basque Country
participated in this experiment: 20 males and 82 females between the age of 21 and 26,
with no previous experience in experiments. They were divided into two groups,
incidental versus intentional, of 51 subjects each. Six subjects from the incidental
group and two from the intentional group were eliminated as a result of improperly
following instructions.
Design
A 2 2 2 2 (Group: Incidental versus Intentional Information: Central versus
Peripheral Content: Actions versus Details Recall trials: First versus Second)
mixed factorial design was used, with Group being the only between-subjects factor.
Following the two recall trials, all subjects responded to the recognition test and rated
their response con®dence.
Material
Subjects received three sequences on a Sony VPH-1000QM Super Bright video projector with a 2.5 2-m screen. First, they heard 2 minutes of news related to drugs.
Afterwards, they saw an emotional ®lm sequence about a kidnapping attempt and,
lastly, they saw 2 minutes of advertisements. The ®lm on the kidnapping lasted 1
minute 10 seconds, with 35 seconds of central information and another 35 seconds of
peripheral information. It begins as a boat arrives at a very busy port and the
passengers get o€. After this peripheral information, a young woman passenger,
following a struggle, is forced into a van, in spite of being helped by another young
man. This central information ends when the girl manages to escape from her kidnappers in a trac jam. The ®lm ®nishes with peripheral information having to do with
a man driving around the port looking for the girl. There is no dialogue in this
sequence, only background music, outdoor noise and the screams that arise from the
situation.
Two independent judges, with experience in incident content and narrative
protocol analysis, evaluated the sequences of the ®lm as central (the kidnapping itself)
and peripheral (the remaining scenes). They also considered actions (directly observable behaviour or facts), determining 35 central actions (e.g. a man lifts up the
tarpaulin of the van or he lunges at the girl) and 35 peripheral actions (e.g. a boat
docks or the passengers get o€). The details, both central and peripheral, were
innumerable and were de®ned (for the purpose of putting together the recognition test
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
260
M. Migueles and E. GarcõÂa-Bajos
and scoring recall) as any descriptive aspect of the ®lm contents (i.e. places, people,
vehicles, objects, and so on).
The recognition test was composed of 24 sentences which followed the event in
chronological order. All the sentences were used to create a coherent narration, adding
the contextual information needed so that each one would make sense. Half of the
sentences were false, though totally plausible in the situation, trying not to include
atypical information. It included six central actions (e.g. true: The young man who
tried to help the girl struggled with the kidnappers; false: Eventually the girl escapes
amid the crowd) and six central details (e.g. true: The hand of one of the kidnappers
was bandaged; false: One of the kidnappers had a moustache). There were also six
peripheral actions (e.g. true: The man who was looking for the girl honked his horn to
clear the way; false: On the dock there were some men loading drums onto a lorry) and
six peripheral details (e.g. true: The name of the boat that docks at the port was called
the Samaina; false: The man looking for the girl drove a white car). Furthermore, an
attempt was made in each case to match the type of contents evaluated, in both actions
and details, in regards to people, vehicles, etc. After each sentence the subjects used a
5-point scale to rate their response con®dence, 1 meaning no certainty and 5 meaning
absolute certainty.
Procedure
The subjects in the incidental group were called to carry out an experiment on time
perception. After broadly outlining some of the results obtained in classical experiments conducted in this area, they were told that their job was to judge the duration of
each of the three sequences they were about to see, based on a 15-second pattern
(the duration of which was not known to the subjects), presented by the experimenter
using a chronometer. Following each sequence, they wrote down on a piece of paper
how long it lasted according to the pattern they were given. The subjects in the
intentional group, who attended the session to carry out an experiment on memory,
were told to pay close attention, as they would later be evaluated. After seeing the
video sequences, both groups were asked to write down the content of the ®lm in as
much detail as possible. When this task was ®nished, both groups were given another
5 minutes to repeat the recall account. Finally, all subjects responded to the recognition test by indicating whether each of the sentences was true or false, as well as
scoring their response con®dence.
RESULTS
Recall
Recall was scored by assigning 1 point for each speci®c unique item of information
(central actions, peripheral actions, central details or peripheral details), whether it
was correct or not. For example, the statement `one of the kidnappers was wearing a
black leather jacket' contains three central details, and the sentence `he stopped the
car, ran to the square and looked for the girl' has three peripheral actions. Recall was
corrected by two judges and the limited discrepancies were cleared up by a third
independent judge.
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
Eyewitness testimony
Table 1.
261
Mean number of correctly recalled units
Recall
Actions
Central
Details
Actions
Peripheral
Details
Incidental
First
Second
7.62
8.84
1.04
2.04
4.84
5.13
3.31
7.62
Intentional
First
Second
10.75
12.02
1.84
3.02
6.37
6.75
5.41
8.98
The subjects' correct answers and errors were analysed using two 2 2 2 2
(Group Information Content Recall trials) ANOVAs.
Correct recall
The results are presented in Table 1. The intentional group (M ˆ 27.57) recalled
more units of the event than did the incidental group (M ˆ 20.22), F(1,92) ˆ 34.66,
p 5 0.0001, and more actions (M ˆ 15.58) than details (M ˆ 8.32) were recalled,
F(1,92) ˆ 185.48, p 5 0.0001. Although the information factor was not signi®cant
(central, M ˆ 11.79; peripheral, M ˆ 12.10), the Information Content, F(1,92) ˆ
381.71, p 5 0.0001, and Group Information Content, F(1,92) ˆ 7.99, p ˆ 0.0058,
interactions were signi®cant. In both groups the same pattern of results emerges.
Subjects produced many central actions and few central details, whereas with average
output there were scarcely di€erences between the recall of peripheral actions and
details. Moreover, the disparity between the recall of central actions and details was
intensi®ed in the intentional group. As a result of this bias in reporting central actions
at the expense of central details, more central than peripheral actions, and more
peripheral than central details were recalled (p 5 0.0001).
In the second trial the subjects recalled more units (M ˆ 27.20) than in the ®rst
(M ˆ 20.59), F(1,92) ˆ 176.32, p 5 0.0001, and the Information Trials, F(1,92) ˆ
15, p ˆ 0.0002, Content Trials, F(1,92) ˆ 33.47, p 5 0.0001, and Information Content Trials, F(1,92) ˆ 55.73, p 5 0.0001, interactions were also signi®cant. The
reason for all of these e€ects stems from the important increase in the recall of
peripheral details, which, in the second trial, was even greater than the recall of
peripheral actions.
Errors
There were very few errors (M ˆ 0.92); 33% of the subjects in the intentional group
and 40% in the incidental group had none. More errors were made in central
information (M ˆ 0.60) than in peripheral information (M ˆ 0.32), F(1,92) ˆ 14.77,
p ˆ 0.0002, in details (M ˆ 0.80) than in actions (M ˆ 0.12), F(1,92) ˆ 51.19,
p 5 0.0001, and in the second recall trial (M ˆ 1.15) than in the ®rst one
(M ˆ 0.69), F(1,92) ˆ 23.23, p 5 0.0001. The Group Content, F(1,92) ˆ 4.95,
p ˆ 0.028, and Content Trials, F(1,92) ˆ 16.29, p ˆ 0.0001, interactions were also
signi®cant. The low rate of error in actions was maintained in groups as well as in
recall trials, while errors in details increased from the ®rst (M ˆ 0.59) to the second
trial(M ˆ 0.99), and the intentional group (M ˆ 0.95) had more errors than the
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
262
Table 2.
M. Migueles and E. GarcõÂa-Bajos
Mean hit rates, false alarm rates, and A0 scores in the recognition task
Actions
Central
Details
Peripheral
Actions
Details
Hits
Incidental
Intentional
0.73
0.78
0.44
0.39
0.58
0.65
0.55
0.50
FAs
Incidental
Intentional
0.85
0.86
0.46
0.46
0.40
0.37
0.53
0.49
A0 s
Incidental
Intentional
0.14
0.18
0.31
0.29
0.48
0.60
0.36
0.23
incidental group (M ˆ 0.63). Almost 50% of the errors in details were related to the
colour of vehicles, clothing and hair.
Recognition
Hit rates, false alarm rates and A0 scores (see Table 2) were analysed by means of three
2 2 2 (Group Information Content) ANOVAs. A0 is a non-parametric analog
to d 0 and is used when d0 cannot be computed because some subjects may produce hit
rates of 1, false alarm rates of 0, both, or the contrary (see the formulas in Snodgrass
et al., 1985). With A0 scores, 0.50 represents chance performance, and higher scores
re¯ect increased sensitivity and accuracy.
Hit rates
The proportion of hits was higher for actions (M ˆ 0.69) than for details (M ˆ 0.47),
F(1,92) ˆ 71.10, p 5 0.0001. There were two interactions: Group Content,
F(1,92) ˆ 4.68, p ˆ 0.033, because the intentional group recognized more actions
and fewer details than did the incidental group; and Information Content,
F(1,92) ˆ 18.95, p 5 0.0001, due to the fact that, following the same pattern of
recall results, in central information there were more hits for actions and fewer for
detail than in peripheral information.
False alarm rates
There was a greater proportion of false alarms in central information (M ˆ 0.66) than
in peripheral information (M ˆ 0.45), F(1,92) ˆ 51.79, p 5 0.0001, and in actions
(M ˆ 0.62) than in details (M ˆ 0.48), F(1,92) ˆ 30.22, p 5 0.0001. The Information Content interaction was also signi®cant, F(1,92) ˆ 117.51, p 5 0.00001, with
the proportion of false alarms in central actions being much higher than in central
details or either sort of peripheral content.
A0 scores
Due to the high proportion of false alarms in central actions, accuracy was greater
in peripheral information (M ˆ 0.42) than in central information (M ˆ 0.23),
F(1,92) ˆ 21.24, p 5 0.00001. There were also two interactions: Group Content,
F(1,92) ˆ 6.42, p ˆ 0.013, because the intentional group was more accurate in actions
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
Eyewitness testimony
263
and less in details than was the incidental group; and Information Content,
F(1,92) ˆ 29.71, p 5 0.00001, re¯ecting equal accuracy in central and peripheral
details, with greater accuracy in peripheral actions and less in central actions.
Con®dence
To begin, con®dence results were analysed following the comparisons generally used
in eyewitness memory research. In this way, the subjects placed greater con®dence on
correct answers (M ˆ 2.57) than on incorrect ones (M ˆ 2.36), F(1,92) ˆ 15.98,
p 5 0.001; and `yes' answers (M ˆ 2.91) were given greater con®dence scores than
were `no' answers (M ˆ 2.01), F(1,92) ˆ 147.36, p 5 0.001. In an actual crime we
would only have been able to carry out the second comparison, as it would have been
impossible to verify the accuracy of the answers. However, simulating an event allows
us to determine the nature of the responses: Hits, FAs, CRs and Misses, and to
analyse con®dence by using the type of responses as a classifying variable.
Average con®dence scores are shown in Table 3. As can be seen, not all subjects
gave responses in all of the categories; this is re¯ected in the degree of freedom of the
variance analyses.
Con®dence was higher in the intentional group (M ˆ 2.72) than in the incidental
group (M ˆ 2.36), F(1,92) ˆ 10.34, p ˆ 0.0018, in central information (M ˆ 3.06)
than in peripheral information (M ˆ 2.03), F(1,92) ˆ 243.38, p 5 0.001, and in actions
(M ˆ 2.82) than in details (M ˆ 2.28), F(1,92) ˆ 103.6, p 5 0.001. The Information Content interaction was signi®cant, F(1,92) ˆ 72.6, p 5 0.001; subjects gave the
same con®dence scores for peripheral actions as they did for details (M ˆ 2.05 and 2.01
respectively), but gave higher con®dence scores for central actions (M ˆ 3.58) than
for central details (M ˆ 2.55), with the response pattern being similar to that of recall
and hits. The Group Information Content interaction was also signi®cant,
F(1,92) ˆ 4.5, p ˆ 0.036. The intentional group had a slightly higher con®dence
rating in central actions and peripheral details than did the incidental group.
In the same way, the response factor was signi®cant, F(3,270) ˆ 71.42, p 5 0.001.
Con®dence was higher in hits (M ˆ 3.02) than in false alarms (M ˆ 2.83), correct
rejections (M ˆ 2.15) and misses (M ˆ 1.89); it was higher in false alarms than in
Table 3. Con®dence means (range 1±5) and number of subjects (in parentheses) in the
recognition task
Central
Details
Peripheral
Actions
Details
Responses
Actions
Incidental
Hits
FAs
CRs
Misses
3.45
3.83
3.16
2.38
(45)
(45)
(16)
(30)
3.30
2.56
2.1
1.72
(42)
(36)
(41)
(42)
2.19
2.01
2.15
1.45
(39)
(35)
(43)
(35)
2.01
1.97
1.43
1.62
(41)
(43)
(41)
(34)
Intentional
Hits
FAs
CRs
Misses
4.19
4.10
3.26
2.53
(49)
(49)
(15)
(27)
3.73
2.85
2.29
2.06
(42)
(41)
(44)
(47)
2.29
2.32
2.17
1.66
(46)
(36)
(46)
(37)
2.60
2.25
2.08
2.04
(42)
(42)
(41)
(43)
Note: Incidental, n ˆ 45; Intentional, n ˆ 49.
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
264
M. Migueles and E. GarcõÂa-Bajos
correct rejections and misses; and higher in correct rejections than in misses. Type of
response interacted with information, F(3,237) ˆ 21.98, p 5 0.001. In the CRs and
misses (`no' answers) con®dence in the central information was only slightly higher
than in peripheral information (M ˆ 0.27 and 0.28 higher respectively), but in the hits
and FAs (`yes' answers) it was much greater (M ˆ 1.34 and 1.23 higher respectively).
The type of response also interacted with the content, F(3,222) ˆ 7.72, p 5 0.001. In
hits, CRs and misses, con®dence in actions was only slightly higher than it was in
details (M ˆ 0.20, 0.40 and 0.10 higher respectively) but in the FAs it increased to a
greater extent (M ˆ 0.86 higher), reaching an even higher level than with hits.
DISCUSSION
Several aspects set this study apart from others that analyse the di€erence between the
central and peripheral information in an emotional event. First, in spite of great
e€orts made for subjects to be incidental observers of the emotional material (e.g.
Burke et al., 1992; Christianson, 1984; Heuer and Reisberg, 1990), there was nothing
which contrasted in a direct and simple fashion the di€erences between incidental and
intentional processing. Second, by using a ®lm instead of static materials, such as
slides, the di€erence between the type of information (central and peripheral) and
evaluated contents (actions and details) was systematically distinguished, thereby
analysing the e€ect of both variables in two successive recall tests, recognition and
response con®dence.
Subjects recalled the same amount of central information (related to the emotional
source) as they did peripheral information (spatially and temporally separated from
the emotional source), showing that the same amount of resources were present during
di€erent moments of the event. But attention was directed to di€erent contents, thus
revealing a clear recall pattern. Whereas in peripheral information there is a balance
between the recall of actions and details, central information is represented basically
by actions to the detriment of more particular details, a pattern which is even more
marked in the intentional group. One might think that the subjects reconstructed the
event by recalling typical actions of the kidnapping script, which would generate
errors; however, as in previous studies (Geiselman et al., 1985; Sanders and Chiu, 1988;
Yuille and Cutshall, 1986), there were very few errors in central actions.
Would the initial disadvantage in central details be diminished if subjects were
given the opportunity to repeat and complete the recall account? With short intervals
between trials, recall improves (Wheeler and Roediger, 1992) and narrations become
longer (Dunning and Stern, 1992; Scrivner and Safer, 1988), but in this study recall
did not increase equally for all contents, improving only slightly in regard to central
details. The ®rst recall revolved around actions and in the second trial subjects
repeated the narration, ®lling it in mainly with peripheral details. As in the work of
Eugenio et al. (1982), errors increased somewhat, basically a€ecting details, and, more
so in central than in peripheral information. Moreover, the level of attention paid to
the event did not seem to be responsible for the e€ect, as the intentional group had
more errors in details than did the incidental group.
Free recall tasks tend to render accurate statements (e.g. Loftus, 1979; Lipton
1977). Thus, in this study subjects showed good recall of the event, with inaccuracies
only having to do with particular details, much the same as in previous studies
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
Eyewitness testimony
265
(Christianson and HuÈbinette, 1993; Yuille and Cutshall, 1986). However, is subjects'
recall a good control index of an event? The recognition test tried to determine this
aspect. If we focus on hits, the recall pattern reappears showing a greater proportion
of correct answers in central actions than in central details and with a balance in the
peripheral information. But the analysis of false alarms and of A's o€ers a very
di€erent perspective. The subjects erroneously accepted 85% of the false central
actions and between 40±50% of the rest of the contents, placing overall accuracy for
the A0 scores below chance performance.
Why do subjects accept false information with such high probability, especially in
central actions? The false recognition sentences, unlike in other studies where
distracters are easier to distinguish (e.g. Burke et al., 1992; Heuer and Reisberg, 1990),
included contents that ®t in well with the event and, as in studies analysing the
misinformation e€ect (e.g. Loftus and Ho€man, 1989; Weingardt et al., 1994),
subjects could have been misled by suggested information. It is also possible that the
subjects based their answers, in part, on the kidnapping script, fundamentally
organised around actions (see Holst and Pezdek, 1992), and that particular details
and more peripheral contents were not as directly speci®ed in the script. The typicality
of the distracters might explain the surprising high rate of false alarms found in
central actions. In an experiment carried out in our laboratory (Migueles and GarcõÂ aBajos, 1996), based on script information established in a previous normative data
study, a brief narrative account of a mock mugging event was elaborated. Systematically, manipulating the typicality of the sentences in a true/false recognition test,
the subjects falsely recognized almost half of the typical sentences not presented in the
narration with a high degree of con®dence, as is the case in the present study, but
scarcely made errors in low-typicality sentences. The practical implications of these
consistent results are evident.
How is recall, recognition and con®dence a€ected by the manipulation of
intentionality? Intentional processing does not introduce major changes in regard to
incidental learning; it merely intensi®es or accentuates the e€ects. Thus, the
group content interactions showed that the intentional group recalled and recognised the actions in the ®lm with greater precision and con®dence, but as far as details
were concerned it did worse than the incidental group. Furthermore, intentionality did
not change the marked tendency to accept false information coherent to the situation.
Con®dence results depended on the type and content of the information as much as
on the nature of the response. For contents, a similar pattern to that of recall emerged
once again: equal con®dence in peripheral actions and details, and greater con®dence
in central actions than in central details. Therefore, con®dence values also ®t in quite
well with the idea of a di€erential distribution pattern of attention for the diverse
contents of the event. In regard to the type of response, it might seem that con®dence
predicts accuracy, since the subjects gave greater con®dence scores to their correct
responses (Hits, CRs) than to their incorrect ones (FAs, Misses), but a high score in
con®dence does not imply that the response is correct (e.g. the FAs). Moreover, as in
studies carried out by Malpass and Devine (1981) on eyewitness identi®cation,
subjects placed more con®dence on their `yes' answers (Hits and FAs) than they did
on their `no' answers (CRs and Misses), regardless of accuracy, rearming the
direction of their decision with their con®dence scores. In any case, the fact that
con®dence was greater in Hits than it was in Misses might indicate that the subjects
had a certain amount of command of the information seen in the ®lm of the
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
266
M. Migueles and E. GarcõÂa-Bajos
kidnapping, but did not distinguish the false information, since the FAs received
greater con®dence scores than did the CRs.
To sum up, when the material allows for precise distinction between the actions and
details of an emotional event, it is noted that eyewitness' attention is drawn to central
actions to the detriment of descriptive details, while as for information not connected
to the emotional focal point, as may be the case of more neutral events, attention is
distributed more homogeneously between peripheral actions and details. As a consequence, more central actions are recalled and recognized than peripheral ones, but
also more peripheral details than central. This latter aspect could be interpreted as
contradictory to the results from previous studies. However, the discrepancies that
strike us at ®rst are only apparent due to the fact that in many cases contents have
been mixed, and in others the materials are so simple (e.g. slides) that the divergence
in results merely expresses how our cognitive system operates in each situation. From
this perspective, the next step is to systematically study the memory pattern of
eyewitnesses for actions and details using di€erent materials and situations, and
tackling cases where central and peripheral information coincides in space and time in
the same scene.
ACKNOWLEDGEMENTS
This research was supported by Grant UPV 227.231-EA133/92 from the University of
the Basque Country. We would like to thank two anonymous reviewers for their
helpful comments on the paper.
REFERENCES
Burke, A., Heuer, F. and Reisberg, D. (1992). Remembering emotional events. Memory and
Cognition, 20, 277±290.
Christianson, S.-AÊ. (1984). The relationship between induced emotional arousal and amnesia.
Scandinavian Journal of Psychology, 25, 147±160.
Christianson, S.-AÊ. (1992). Emotional stress and eyewitness memory: A critical review.
Psychological Bulletin, 112, 284±309.
Christianson, S.-AÊ., Forth, A. E., Hare, R. D., Strachan, C., Lidberg, L. and Thorell, L. H.
(1996). Remembering details of emotional events: A comparison between psychopathic and
nonpsychopathic o€enders. Personality & Individual Di€erences, 20, 437±443.
Christianson, S.-AÊ. and HuÈbinette, B. (1993). Hands ups! A study of witnesses' emotional
reactions and memories associated with bank robberies. Applied Cognitive Psychology, 7,
365±379.
Christianson, S.-AÊ. and Loftus, E. F. (1987). Memory for traumatic events. Applied Cognitive
Psychology, 1, 225±239.
Christianson, S.-AÊ. and Loftus, E. F. (1990). Some characteristics of people's traumatic
memories. Bulletin of the Psychonomic Society, 28, 195±198.
Christianson, S.-AÊ. and Loftus, E. F. (1991). Remembering emotional events: The fate of
detailed information. Cognition and Emotion, 5, 81±108.
Christianson, S.-AÊ. and Nilsson, L. G. (1989). Hysterical amnesia: A case of aversively
motivated isolation of memory. In T. Archer and L. G. Nilsson (Eds), Aversion, avoidance,
and anxiety: Perspectives on aversively motivated behavior. Hillsdale, NJ: Erlbaum.
Cli€ord, B. R. and Hollin, C. R. (1981). E€ects of the type of incident and the number of
perpetrators on eyewitness memory. Journal of Applied Psychology, 66, 364±370.
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
Eyewitness testimony
267
Cli€ord, B. R. and Scott, J. (1978). Individual and situational factors in eyewitness testimony.
Journal of Applied Psychology, 63, 352±359.
Cutshall, J. L. and Yuille, J. C. (1989). Field studies of eyewitness memory of actual crimes.
In D. C. Raskin (Ed.), Psychological methods in criminal investigation and evidence. New
York: Springer.
Dunning, D. and Stern, L. B. (1992). Examining the generality of eyewitness hypermnesia:
A close look at time delay and question type. Applied Cognitive Psychology, 6, 643±657.
Easterbrook, J. A. (1959). The e€ect of emotion on cue utilization and the organization of
behavior. Psychological Review, 66, 183±201.
Eugenio, P., Buckhout, R., Kostes, S. and Ellison, K. (1982). Hypermnesia in the eyewitness to
a crime. Bulletin of the Psychonomic Society, 19, 83±86.
Geiselman, R. E., Fisher, R. P., MacKinnon, D. P. and Holland, H. L. (1985). Eyewitness
memory enhancement in the police interview: Cognitive retrieval mnemonics versus
hypnosis. Journal of Applied Psychology, 70, 401±412.
Heuer, F. and Reisberg, D. (1990). Vivid memories of emotional events: The accuracy of
remembered minutae. Memory & Cognition, 18, 496±506.
Heuer, F. and Reisberg, D. (1992). Emotion, arousal, and memory for detail. In S.-AÊ.
Christianson (Ed.), The Handbook of emotion and memory: Research and Theory. Hillsdale,
NJ: Erlbaum.
Holst, V. F. and Pezdek, K. (1992). Scripts for typical crimes and their e€ects on memory for
eyewitness testimony. Applied Cognitive Psychology, 6, 573±588.
Kebeck, G. and Lohaus, A. (1986). E€ect of emotional arousal on free recall of complex
material. Perceptual and Motor Skills, 63, 461±462.
Kuehn, L. L. (1974). Looking down a gun barrel: Person description and violent crime.
Perceptual and Motor Skills, 39, 1159±1164.
Lipton, J. P. (1977). On the psychology of eyewitness testimony. Journal of Applied Psychology,
62, 90±95.
Loftus, E. F. (1979). Eyewitness testimony. London: Harvard University Press.
Loftus, E. F. and Burns, T. (1982). Mental shock can produce retrograde amnesia. Memory &
Cognition, 10, 318±323.
Loftus, E. F. and Ho€man, H. (1989). Misinformation and memory: The creation of new
memories. Journal of Experimental Psychology: General, 118, 100±104.
Malpass, R. S. and Devine, P. G. (1981). Eyewitness identi®cation: Lineup instructions and the
absence of the o€ender. Journal of Applied Psychology, 66, 482±489.
Migueles, M. and GarcõÂ a-Bajos, E. (1996). The role of scrips in the accuracy±con®dence
relationship: Implications for eyewitness memory. The Second International Conference on
Memory. Padova, Italy.
Reisberg, G., Heuer, F., McLean, J. and O'Shaughnessy, M. (1988). The quantity, not the
quality, of a€ect predicts memory vividness. Bulletin of the Psychonomic Society, 26,
100±103.
Sanders, G. S. and Chiu, W. (1988). Eyewitness errors in the free recall of actions. Journal of
Applied Social Psychology, 18, 1241±1259.
Scrivner, E. and Safer, M. A. (1988). Eyewitnesses show hypermnesia for details about a
violent event. Journal of Applied Psychology, 73, 371±377.
Snodgrass, J. G., Levy-Berger, G. and Haydon, M. (1985). Human experimental psychology.
New York: Oxford University Press.
Turtle, J. W. and Yuille, J. C. (1994). Lost but not forgotten details: Repeated eyewitness
recall leads to reminiscence but not hypermnesia. Journal of Applied Psychology, 79,
260±271.
Wagenaar, W. A. and Groeneweg, J. (1990). The memory of concentration camp survivors.
Applied Cognitive Psychology, 4, 77±87.
Weingardt, K. R., Toland, H. K. and Loftus, E. F. (1994). Reports of suggested memories:
Do people truly believe them? In D. F. Ross, J. D. Read and M. P. Toglia (Eds), Adult
eyewitness testimony. Current trends and developments. New York: Cambridge University
Press.
Wessel, I. and Merckelbach, H. (1994). Characteristics of traumatic memories in normal
subjects. Behavioral and Cognitive Psychotherapy, 22, 315±324.
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)
268
M. Migueles and E. GarcõÂa-Bajos
Wessel, I. and Merckelbach, H. (1997). The impact of anxiety on memory for details in spider
phobics. Applied Cognitive Psychology, 11, 223±231.
Wheeler, M. A. and Roediger, H. L. III (1992). Disparate e€ects of repeated testing:
Reconciling Ballard's (1913) and Bartlett's (1932) results. Psychological Science, 3, 240±245.
Yuille, J. C. and Cutshall, J. L. (1986). A case study of eyewitness memory of a crime. Journal
of Applied Psychology, 71, 291±301.
Copyright # 1999 John Wiley & Sons, Ltd.
Appl. Cognit. Psychol. 13: 257±268 (1999)