SOCIAL NEUROSCIENCE, 2018

VOL. 13, NO. 1, 74–93

https://doi.org/10.1080/17470919.2016.1253610

Are eyewitness accounts biased? Evaluating false memories for crimes involving
in-group or out-group conflict
a
Alexis C. Carpenter and Anne C. Krendlb
a
Department of Psychology, Harvard University, Cambridge, MA, USA; bDepartment of Psychological & Brain Sciences, Indiana University,
Bloomington, IN, USA

ABSTRACT ARTICLE HISTORY

Eyewitness testimony has been shown to be unreliable and susceptible to false memories. Received 19 June 2016
Whether eyewitness memory errors are influenced by the victim’s group membership (relative Revised 19 October 2016
Published online 16
to both the eyewitness and perpetrator) is underexplored. The current study used complemen- November 2016
tary behavioral and neuroimaging approaches to test the hypothesis that intragroup conflict
heightens participants’ susceptibility to subsequent false memories. Healthy young adults wit- KEYWORDS
nessed and later answered questions about events in which the perpetrator and victim were False memory; episodic
either 1) identified as in-group members relative to each other and the eyewitness, 2) out-group memory; intragroup conflict;
members relative to the eyewitness, but not each other, or 3) out-group members relative to social salience; fMRI
each other (Experiments 1a and 1b). When perpetrators and victims were in-group members
(intragroup conflict), participants showed heightened false memory rates. Moreover, false mem-
ories increased upon crime realization. Neuroimaging data analysis revealed that salient (as
compared to ambiguous) intragroup conflict elicited heightened activation in neural regions
associated with resolving cognitive conflict (anterior cingulate cortex; ACC). Increased functional
connectivity between the ACC and dorsomedial prefrontal cortex was associated with subse-
quent false memories (Experiment 2). Results suggest that the social salience of the intragroup
conflict may have been associated with participants’ increased susceptibility to false memories.

Although juries generally weigh eyewitness testimony memories (but see Nourkova, Bernstein, & Loftus, 2004).
heavily in their deliberations (Abshire & Bornstein, 2003; The current studies examined these questions.
Lindsay, Lim, Marando, & Cully, 1986), eyewitness accounts Prior work demonstrated the importance of tying event
are often unreliable and easily susceptible to false mem- details to their context or source to support true memories,
ories (Loftus, 1975, 1979, 2005; Loftus, Miller, & Burns, 1978; particularly for social information (Howard & Rothbart,
Mitchell & Zaragoza, 2001; Okado & Stark, 2005; Zaragoza, 1980). According to this work, false memories for crime-
Belli, & Payment, 2006; Zaragoza & Lane, 1994). Much related details arise from mistakenly combining elements of
research on false memories in eyewitness testimony has past experiences (i.e., source misattributions; Okado & Stark,
focused on memory errors regarding the identity of the 2005; Stark, Okado, & Loftus, 2010). In an experimental
perpetrator of the crime (e.g., Morgan III, Southwick, context, these source misattributions may emerge when
Steffian, Hazlett & Loftus, 2013), including how the perpe- participants retrieve incorrect information about the event
trator’s group membership (relative to the eyewitness) may they witnessed, and attribute their memory for that infor-
influence memory errors (e.g., Howard & Rothbart, 1980). mation to the original event. Indeed, memory errors can
However, a small body of research suggests that eyewit- result from mistakenly combining details of distinct episo-
nesses’ false memories may also extend to the details of the dic or autobiographical memories (e.g., Burt, Kemp, &
crime itself (e.g., Loftus, 2005; Okado & Stark, 2005). Two Conway, 2004; Carpenter & Schacter, in press; Devitt,
critical and overlooked considerations in this work regard: Monk-Fromont, Schacter, & Addis, 2016; Odegard &
1) whether these factors interact, and 2) how a victim’s Lampinen, 2004). Eyewitnesses were particularly prone to
group membership (relative to both the eyewitness and these types of source monitoring errors following a time
perpetrator) impacts eyewitnesses’ susceptibility to false delay, during which memory for event-related details

CONTACT Alexis C. Carpenter [email protected] Department of Psychology, Harvard University, 33 Kirkland St., Cambridge, MA 02138,
USA
Supplemental data for this article can be accessed here.
© 2016 Informa UK Limited, trading as Taylor & Francis Group
 SOCIAL NEUROSCIENCE 75

deemed to be behaviorally irrelevant may weaken or fade If both the victim and the perpetrator are in-group
(Hardt, Nader, & Nadel, 2013). During this delay, exposure to members (relative to each other and the witness), two
alternate sources of information related to details of the divergent possibilities emerge. First, fewer source misattri-
witnessed event (e.g., accounts of other witnesses, news butions may occur when eyewitnesses view an event
reports, and even the questions asked by the investigator) depicting two in-group members relative to the in-group
may incorrectly modify witnesses’ memory traces. Given victim condition described above because these crimes
weak original traces, individuals may not detect discrepan- may have the most salience to eyewitnesses by involving
cies between reactivated event details and post-event mis- two in-group members versus one. Second, eyewitnesses
information contradicting original event details. As a result, may not attend as much to crime-related details with both
witnesses may misattribute the source of the more recently in-group perpetrators and victims because the intragroup
presented misinformation to the original event, creating conflict may violate behavioral expectations of in-group
false memories (e.g., Hupbach, Gomez, & Nadel, 2009; see members. Indeed, people have worse memory for negative
Mitchell & Johnson, 2009; for a review). in-group versus out-group member behaviors, suggesting
The extent to which the eyewitnesses attend to the preferential memory for expectation-consistent behaviors
salient elements of the event (e.g., the victim and/or perpe- (Howard & Rothbart, 1980). Here, intragroup conflict could
trator of a crime), at the expense of attending to other, less result in more false memories than conflict between two
salient elements of an event (e.g., the item that was stolen), out-group members relative to the witness, ambiguous
may influence the strength of original memory traces. For conflict (e.g., an in-group member victim relative to the
instance, the weapon-focus effect (Loftus, 1979) demon- witness, but a perpetrator from an unidentified group), or
strates that people remember central emotional elements intergroup conflict (e.g., in-group member victim relative to
of events (e.g., the weapon) at the expense of other details the witness, but out-group member perpetrator relative to
(e.g., peripheral elements within the scene). We therefore both).
extended eyewitness false memory research to the domain To resolve these disparate possibilities, we assessed eye-
of group membership, which we anticipated would have witnesses’ false memory for crimes involving either in-
high social salience for the eyewitnesses. We anticipated group victims and in-group perpetrators or out-group vic-
that victim and/or perpetrator group membership (relative tims and out-group perpetrators (Experiment 1a). We then
to the eyewitness and/or each other) would affect eyewit- expanded these findings in Experiment 1b by examining
nesses’ source misattributions for event-related details by false memory rates for intergroup conflict (in-group victim,
drawing their attention away from less salient crime-related out-group perpetrator). To better characterize our pre-
details and toward the more salient conflict between the dicted results, we used functional magnetic resonance
victim and perpetrator. There are several ways in which this imaging (fMRI) to identify the neural mechanisms under-
could occur. lying source misattributions for crime-related details
For instance, source misattributions could be lower impacted by victim and/or perpetrator group membership
for crimes involving victims who are in-group members (Experiment 2).
relative to the witness (irrespective of the perpetrator’s Across all experiments, we used university affiliation
group membership) because the eyewitnesses may (in-group: Indiana University, out-group: Purdue
view an in-group victim as being more similar to them- University) to manipulate group membership (e.g., see
selves than an out-group victim. If this is the case, then Bernstein, Young, & Hugenberg, 2007; Wilder, 1990).
eyewitnesses may form stronger memory traces for
details of the original event. Relating information to
Experiment 1a
the self yields memory enhancements because self-
reference promotes greater elaboration and organiza- Here, we examined whether the victim and/or perpetrator
tion of information during encoding (see Symons & group membership relative to the witness affected wit-
Johnson, 1997 for a meta-analysis). Supporting this pre- nesses’ false memories for crime-related details. We pre-
diction, individuals affiliate less with in-group members dicted that false memories would differ for crimes depicting
violating social norms (e.g., committing a crime), but an in-group victim and an in-group perpetrator. However,
more when they are victims of social norm violations we tested two competing hypotheses regarding the direc-
(Eidelman & Biernat, 2003; Marcus-Newhall, Blake, & tionality of this effect. If self-relevance improves memory for
Baumann, 2002; Marques & Yzerbyt, 1988; Nourkova crime-related details, we predicted fewer false memories in
et al., 2004; Pinto, Marques, Levine, & Abrams, 2010). a crime involving any in-group member (victim and/or
Thus, stronger affiliation with the in-group victim may perpetrator) as compared with the out-group condition. If
boost memory performance because the in-group vic- intragroup conflict (shared in-group membership between
tim would be seen as being similar to the self. victim, perpetrator, and eyewitness) directs attention
76 A. C. CARPENTER AND A. C. KRENDL

toward resolving behavioral conflict between the in-group received course credit or monetary compensation. The
members, we anticipated more false memories given Indiana University IRB approved this research.
intragroup conflict only. To disentangle the effect of a
crime featuring an in-group victim from a crime featuring
an in-group victim and in-group perpetrator, we presented Encoding the original events
participants with crimes in which the conflict was either
Participants viewed a modified version of slideshows used
ambiguous (e.g., the victim was explicitly identified as
by Okado and Stark (2005) that depicted (in counterba-
being an in-group or out-group member, but the perpe-
lanced order) eight unique everyday life events. Six events
trator-s group membership was not explicitly identified), or
depicted criminal activities (e.g., a wallet being stolen), and
salient (e.g., the victim and the perpetrator were both
two depicted neutral activities (e.g., a woman going to the
explicitly identified as being in-group or out-group
grocery store). Each slideshow contained 50 images, pre-
members).
sented for 3500 ms each with a 500 ms inter-picture interval
(e.g., black screen). Participants were told that they would
rate the slideshows on a variety of measures in a second
Materials and methods session (approximately 24-hrs later), so it was important
that they were attending to the images presented, but
Participants
were unaware that they would have a future memory test.
A priori power analyses conducted using G*Power (Faul, In the six events depicting criminal activity, there was a
Erdfelder, Buchner, & Lang, 2009) indicated at least 26 clearly identifiable perpetrator and a victim. To manipulate
participants would be necessary per condition to detect perceived group membership of the perpetrator and/or the
within-subject effects using a medium effect size (ηp2 = .10), victim, we used Adobe Photoshop Elements 11 Version 11.0
alpha = .05, and power = .95. Because participants were to superimpose symbols (e.g., logos, banners, or building
required to complete two sessions and needed to achieve names) explicitly associated with either Indiana University
above chance performance on the memory test to be (in-group) or Purdue University (out-group) onto each of
included, a total of 53 undergraduates (Mage = 22 years, the 50 images for each of the eight slideshows (see example
SD = 2.96, 38 female, all White) from Indiana University with in Figure 1). We chose Purdue University as the out-group
normal or corrected-to-normal vision and hearing, who because Indiana University (where the study was con-
were not taking antidepressants, and who were native ducted) and Purdue University are both public universities
English speakers, were recruited. Participants either in Indiana, and there is a well-known rivalry between the

TARGET
Victim & Perpetrator Victim Only
Indiana University (IG)
SCHOOL
Purdue University (OG)

Figure 1. An example image from one slideshow depicting how images were manipulated to identify the victim and the
perpetrator (i.e., salient conflict condition) or only the victim (i.e., ambiguous conflict condition) as an in-group (i.e., Indiana
University) or an out-group (i.e., Purdue University) member. In each condition, the same items were manipulated (e.g.,
perpetrator’s jacket, victim’s wallet, victim’s mouse pad) while information not directly linked to either the victim or the perpetrator
was held constant (e.g., background items). Manipulations circled in blue identify the victim and manipulations circled in red
identify the perpetrator as either in-group or out-group members. Participants saw these images without the circles.
 SOCIAL NEUROSCIENCE 77

two universities among the students. A posttest manipula- (Indiana University or Purdue University) was manipulated
tion check verified the validity of our in-group/out-group between subjects. Conflict conditions were presented in
identification measures. Participants were asked two ques- pseudorandom order across participants.
tions: 1) How important is being a part of Indiana University
to you? and 2) How important is being a part of Purdue
Introducing the misinformation
University to you? Participants both affiliated significantly
more with Indiana University as compared to Purdue Approximately 24-hrs after viewing the original slideshows,
University, and were able to successfully link each identified participants listened to an audio-recording for each of the
victim and perpetrator to their respective university. eight events to ostensibly remind them of the events they
We manipulated the images to create two ambiguous had seen the day prior (see Figure 2; see supplemental
group conflict conditions: an in-group-victim ambiguous Figure 1). Each recording consisted of one sentence describ-
conflict condition (in which the victim was identified as ing each of the slide images that participants had seen the
being an in-group member, but the perpetrator’s group day prior, totaling 50 sentences for each of the eight events.
membership was not identified), and an out-group-victim The events were presented in pseudorandom order.
ambiguous conflict condition (in which the victim was Critically, 12 of the 50 sentences for each event were incon-
identified as being an out-group member, but the perpe- sistent with what had been presented in the original slide-
trator’s group membership was not identified). We also shows (i.e., introducing misinformation), whereas the
had two salient group conflict conditions: a salient in- remaining 38 sentences were consistent with the slide-
group conflict condition (in which both the victim and show). Consistent with previous research, the misinforma-
the perpetrator were both identified as being in-group tion was not supplementary; it directly contradicted
members), and a salient out-group conflict condition (in originally presented information (Lindsay & Johnson,
which both the victim and the perpetrator were both 1989; Loftus et al., 1978; Okado & Stark, 2005; Zaragoza &
identified as being out-group members). Thus, four con- Lane, 1994). Each sentence was read at a comparable pace
ditions of interest were created: (1) Salient in-group con- (averaging 1 sentence every 4 s). Participants listened to the
flict: both the victim and the perpetrator were members recordings using high-quality headphones, and all verified
of the in-group (IU); (2) Salient out-group conflict: both to the experimenter that they could hear the recordings
the victim and the perpetrator were members of the out- after the first trial.
group (PU); (3) In-group-victim ambiguous conflict: the
victim was an in-group member (IU) and the perpetrator
Measuring true and false memories
was unidentified; (4) Out-group-victim ambiguous conflict:
the victim was an out-group member (PU) and the perpe- After listening to all eight audio-recordings, we asked parti-
trator was unidentified. The exact same items in each cipants 18 forced-choice recognition questions about each
image were modified in the salient in-group and out- of the eight events they had witnessed (based on the
group conflict conditions (see Figure 1). Conflict condition methods by Stark et al., 2010). Participants were instructed
(ambiguous or salient) was manipulated within-subjects to respond based only on their memory for information
(i.e., four slideshows depicted ambiguous group conflict, presented visually in the slideshows 24-hrs prior. Of the 18
and four salient group conflict). Group membership questions, 12 corresponded to the images for which

Figure 2. Order of events in Session 1 (e.g., Slideshow) and Session 2 (e.g., Audio-recording, Memory Test, Source-Monitoring; all of which
were administered 24 hrs after the original encoding in Session 1). Slideshows were viewed during the original encoding in Session 1.
Following a 24-hr delay, participants returned to the laboratory for Session 2. At this time, they heard corresponding audio-recordings,
followed by a multiple choice memory test, and a source-monitoring test. The correct answer is defined as information that participants
saw during the slideshow and is circled in green in the image. The incorrect “misinformation” answer is presented verbally during the
narrative and is presented in red font. A true memory would be defined as choosing the correct answer “A stack of CDs” and attributing
their memory to the slideshow on the source-monitoring test. A false memory would be defined as choosing the “misinformation” answer
“A computer mouse” and incorrectly attribute their memory to the slideshow on the source-monitoring test.
78 A. C. CARPENTER AND A. C. KRENDL

misinformation had been provided, and six control ques- during the audio-recording and attribute the source of their
tions corresponded to the images for which the audio- memory to the slideshow (thereby suggesting that the
recording information was consistent with the slideshow. misinformation from the audio-recording was now incor-
Although none of the test questions addressed group porated into their original memory trace; see Figure 2). To
membership, a post-test manipulation check verified that be considered a true memory, participants had to select the
all participants correctly identified the relevant group mem- answer on the memory accuracy test corresponding to the
berships in each condition. Participants viewed a cut out of slideshow and attribute the source of their memory to the
each individual from each slideshow and were asked: slideshows (see Figure 2). Other memories were those that
“Where was [the victim/perpetrator] from?”. Participants sub- did not meet the above criteria, and will not be discussed
mitted their responses to the open-ended question by further.
typing into the textbox. Participants correctly affiliated
each individual from the slideshows to their corresponding
False memory performance as a function of target
school on 91% (SD = 12.46%) of the trials.
and group membership
Each of the 18 questions was presented with three
possible answer choices. For the misinformation ques- We subjected the number of false memories into a 2
tions, these were: the correct information presented in (conflict type: ambiguous or salient) × 2 (group member-
the slideshows, the misinformation presented in the ship: in-group or out-group) mixed analysis of variance
audio-recordings, and one piece of unrelated informa- (ANOVA). Results revealed a trend for an effect of conflict
tion. For the control questions, these were: the correct type (F(1,49) = 3.28, p = .08, η2partial = .06) and no effect of
information presented in the slideshows (that was also group membership (F(1,49) = 1.27, p = .27, η2partial = .03).
corroborated in the audio-recording), and two distinct The predicted conflict type × group membership inter-
unrelated pieces of information. Upon completion of action emerged, F(1,49) = 4.67, p = .036, η2partial = .09).
the memory test, participants completed a source-mon- Participants had more false memories in the salient in-
itoring test in which they attributed the source of their group conflict condition (MFalse memories = 9.81, SD =
answers from the memory test to the slideshow from 4.07) than the salient out-group conflict condition
session 1, the audio-recording from session 2, or (MFalse memories = 7.52, SD = 4.01; t(49) = 2.02, p = .049;
unknown (see Figure 2; see supplemental Figure 1). 95% CI [.013, 4.56]), the in-group-victim ambiguous group
conflict condition (MFalse memories = 7.54, SD = 3.41; t
(25) = 3.12, p = .004; 95% CI [.77, 3.77]), and trending for
Results
the out-group-victim ambiguous group conflict condition
We excluded two participants (both from the salient in- (MFalse memories = 7.72, SD = 4.11; t(49) = 1.82, p = .074; 95%
group conflict condition) because their memory perfor- CI [-0.21, 4.39]). The other conditions did not differ from
mance on the control questions was below chance each other (all ts < 1).
(33.3%), which suggested poor memory overall. We identi-
fied true and false memories based only on the responses
True memory performance as a function of target
to the 12 misinformation items on the memory accuracy
and group membership
and source-monitoring tests (similar to the approach used
by Carpenter & Schacter, in press; Okado & Stark, 2005; Stark We next entered true memories into the same described
et al., 2010). To be considered a false memory, participants ANOVA. There were no main effects, Fs < 1, ps > .74 and
had to select the answer on the memory accuracy test that no conflict type × group membership interaction, F
corresponded to the misinformation item they had heard (1,49) = 1.39, p = .24. See Table 1 for means.1
1
If the memory test were a simple recognition test—during which we gave participants three options (i.e., correct, misinformation, foil)—
when there is an increase in false memories, there must be a decrease in the number of misses or a decrease in the number of true
memories. However, during the source-monitoring task, participants are given three possible answer choices (i.e., the slideshow, the
audio-recording, or unsure). It is important to note that an increase in false memories does not necessarily mean that the number of
correct memories would decrease. Unlike simple recognition tasks in which participant responses are typically constrained to three
possible response options (i.e., correct, misinformation, foil), there are not only three possible outcomes for source-monitoring
paradigms that measure false memories. Rather, in the source-monitoring paradigm, a false memory is defined as trials for which
participants both choose the misinformation item (out of three possible items) and attribute this information to the original (i.e., visual)
source (out of three possible items), whereas a correct memory is defined as trials for which participants both choose the correct item
(out of three items) and attribute this information to the original source (out of three items). By defining false memories this way (which is
consistent with previous research utilizing the source-monitoring paradigm—for example, Carpenter & Schacter, in press; Okado & Stark,
2005), false memories and correct memories do not necessarily constrain one another in a source memory paradigm as they might in a
recognition paradigm.
 SOCIAL NEUROSCIENCE 79

Table 1. Mean true memory in Experiment 1a and 1b. SD (). Table 2. Mean false memory in Experiment 1a and 1b for the
Indiana University (IG) Purdue University (OG) 1st and 2nd Halves. SD ().
Experiment 1a Indiana University Purdue University
Salient conflict 12.96 (5.85) 13.04 (7.34) (IG) (OG)
Ambiguous conflict 13.77 (5.54) 12.56 (6.77) Experiment 1a
Experiment 1b Salient conflict 1st Half .15 (.10) .12 (.09)
Intragroup conflict 11.50 (4.22) 2nd Half .27 (.12) .18 (.13)
Intergroup conflict 10.99 (4.85) Ambiguous conflict 1st Half .13 (.10) .15 (.13)
IG: In-group, OG: Out-group. 2nd Half .18 (.10) .19 (.10)
Experiment 1b
Intragroup conflict 1st Half .16 (.10)
2nd Half .26 (.11)
Comparing memory errors pre- and post-crime Intergroup conflict 1st Half .19 (.11)
2nd Half .23 (.12)
realization Intragroup conflict = Victim and Perpetrator: Indiana University (IG), Intergroup
Conflict = Victim: Indiana University (IG), Perpetrator: Purdue University (OG).
An important consideration in understanding how vic-
tim and/or perpetrator group membership might affect
eyewitnesses’ false memory rates for crime-related
details is that the roles of victims and perpetrators are group membership (F(1,49) = 6.38, p = .02, η2partial = .12)
not always clearly defined from the beginning of a emerged (see Table 2 for means).
conflict. Thus, only once the conflict between the per- The predicted time × group membership interaction
petrator and victim becomes apparent (which likely also emerged (F(1,49) = 5.10, p = .03, η2partial = .09). There
only occurs well into the crime taking place) would was a higher proportion of false memories post-crime reali-
conflict resolution be necessary, resulting in the pre- zation as compared with pre- for the salient in-group con-
dicted increase in false memory. We therefore also flict condition (MChange in false memories = .12, SD = .12) versus
examined whether false memories changed pre- as the salient out-group conflict condition (MChange in false mem-
compared with post-crime realization across all of our ories = .05, SD = .08; t(49) = 2.26, p = .03; 95% CI [.07, .03],

experiments. If, for instance, intragroup conflict is asso- d = .64). No difference in the proportion of false memories
ciated with increased false memories because the con- between these two conditions occurred in the first half
flict violates eyewitnesses' expectations, then false (salient in-group conflict: MFalse memories = .15, SD = .10; sali-
memories should be higher after the conflict has ent out-group conflict: MFalse memories = .12, SD = .09; t
become salient. (49) = 1.09, p = .28; 95% CI [-.03, .09], d = .30). See Figure 3
Salient in-group versus out-group conflict may have for change in proportion of false memories from the 1st to
elicited more false memories because participants may the 2nd half.
have been distracted by the nature of the intragroup
conflict. If this is the case, the number of false memories
should be exacerbated after the in-group (relative to the
out-group) conflict became salient.
To examine this possibility, we compared the propor-
tion of false memories post-crime realization (the second
half of each event) with pre-crime realization (the first
half) for the in-group and out-group salient conflict con-
ditions. Participant feedback suggested that the halfway
point in each event was when the victim and perpetrator
roles became salient (see supplemental Figure 1). Because
there were not equal numbers of misinformation ques-
tions for the first and second halves of each event, we
calculated a proportion of false memories score by divid- Figure 3. Change in the proportion of false memories from the
ing the number of false memories in the respective half by 1st to the 2nd half for victim and perpetrator versus victim only
the total number of questions asked about misinforma- conditions for both the in-group (IG: Indiana University) and
out-group (OG: Purdue University). Results show significantly
tion items in that half. These data were then entered into a more false memories when both the victim and the perpetrator
2 (time: pre crime-realization or post crime-realization) × 2 are identified as being in-group members during the 2nd half as
(group membership: in-group or out-group) ANOVA. Main compared to the in-group victim-only and both out-group
effects of time (F(1,49) = 33.93, p < .001, η2partial = .41) and conditions. Error bars SEM.
80 A. C. CARPENTER AND A. C. KRENDL

Experiment 1b excluded for failing to comply with task instructions

(e.g., using the phone). No participants had memory
Experiment 1a suggests increased eyewitnesses suscept-
below chance on the six control questions.
ibility to false memories after intragroup conflict became
Experiment 1b replicated Experiment 1a with two
salient. One limitation in that experiment, however, was
differences. First, in order to streamline the task, we
that we did not examine whether salient intergroup con-
only included the six slideshows that depicted criminal
flict (e.g., when the group membership of the victim and
activity (since the two control slideshows were not
perpetrator explicitly differs) also exacerbates false mem-
considered in the analyses anyway). Second, in order
ories. Because heightened perceiver affiliation with a vic-
to compare false memories for salient intragroup and
tim exacerbates perceived conflict, (e.g., Eidelman &
intergroup conflict, the perpetrator’s group member-
Biernat, 2003; Marcus-Newhall et al., 2002; Marques &
ship was either manipulated to be the same as the
Yzerbyt, 1988; Nourkova et al., 2004; Pinto et al., 2010),
victim’s (both in-group members) or different (mixed
we focused on crimes involving an in-group member as
group membership). Conflict was manipulated within
the victim, but manipulated the group membership of the
participant, and the intragroup and intergroup events
perpetrator to be an in-group or out-group member.
were counterbalanced across participants (see Figure 4).
Given our focus on conflict salience, we predicted that
false memories would be exacerbated once conflict was
recognized (e.g., more false memories in the response to Results
information presented in the second half of the events as
compared to the first half). False memory performance as a function of time
and conflict type
As in Experiment 1a, we calculated the proportion of false
Methods
memories pre- and post-crime realization to control for
A priori power analyses conducted using G*Power (Faul the fact that there was different number of critical items
et al., 2009) indicated at least 21 participants would be pre- and post- crime realization (which therefore would
necessary per condition to detect within-subject effects lead to the possibility for a different raw number of false
mirroring the time × group membership effect size from memories overtime). A 2 (conflict type: intragroup or
Experiment 1b (ηp2 = .09), alpha = .05, and power = .95. intergroup) × 2 (time: pre crime-realization or post
Thus, 42 young adults (Mage = 21 years, SD = 2.31, 27 crime-realization) ANOVA on false memory rates revealed
female) from Indiana University were recruited for a main effect of time (F(1,39) = 18.11, p < .001,
Experiment 1b. They met the same eligibility criteria as η2partial = .32;Pre-crime realization: MIntragroup = .16,
outlined in Experiment 1a. Two participants were SD = .10; MIntergroup = .19, SD = .11; Post-crime realization:

TARGET
Ingroup Victim & Perpetrator Mixed Victim & Perpetrator

Figure 4. One of 50 images displayed in one slideshow, manipulated to identify both the victim and the perpetrator as in-group
(i.e., intragroup conflict condition; Indiana University) members (i.e., identical Experiment 1a) or the victim as an in-group (i.e.,
Indiana University) and the perpetrator as an out-group member (i.e., intergroup conflict condition; Purdue University). Items
identifying the perpetrator as an in-group member (the Indiana University logo on his jacket) and the victim as an in-group member
(the computer mousepad and wallet) were held constant throughout the slideshow. Manipulations circled in blue identify the victim
and manipulations circled in red identify the perpetrator as either in-group or out-group members. Participants saw these images
without the circles.
 SOCIAL NEUROSCIENCE 81

Discussion
The results of Experiments 1a and 1b suggest that eye-
witnesses may be more susceptible to false memories for
events depicting intragroup conflict (i.e., when both the
victim and the perpetrator are in-group members).2
Specifically, in Experiment 1a, we discovered that eyewit-
nesses reported more false memories after witnessing a
crime depicting a victim and perpetrator who were in-
group members relative to each other, and the eyewitness,
as compared to when they witnessed a crime depicting two
out-group members, or ambiguous group conflict.
Moreover, the proportion of false memories increased
Figure 5. Change in the proportion of false memories from the post-crime realization as compared pre-crime realization,
1st to the 2nd half for the intragroup conflict versus intergroup suggesting that the salience of the conflict may have been
conflict conditions. Results show significantly more false mem-
ories when both the victim and the perpetrator are identified associated with increased false memories. Experiment 1b
as being in-group members (i.e., intragroup conflict condition) further shed light on these findings by suggesting that
during the 2nd half compared to the mixed victim-perpetrator witnesses experience more false memories post-crime rea-
2nd half. Error bars SEM. lization as compared to pre-crime realization, given crimes
depicting intragroup versus intergroup conflict. Although
MIntragroup = .26, SD = .11; MIntergroup = .23, SD = .12). There consistent with the pattern from Experiment 1a, this rela-
was no main effect of conflict type (F(1,39) < 1, p = .96, tionship was marginally significant. However, it is important
η2partial < .001). to note that both the directionality and effect size of the
Consistent with Experiment 1a, a marginal conflict type interaction in Experiment 1b replicated the results in 1a.
x time interaction emerged, F(1,39) = 3.65, p = .06, Nevertheless, one possibility for the differences in levels of
η2partial = .09). Although the interaction did not meet tradi- significance is that the difference in false memories
tional levels of statistical significance, the directionality and between salient intragroup and intergroup conflict
effect size of these results replicated those in Experiment (Experiment 1b) may be weaker than the effects observed
1a. Specifically, the interaction was driven by the fact that between ambiguous and salient group conflict
there was a larger increase in the proportion of false mem- (Experiment 1a).
ories from pre- to post- crime realization for the intragroup Several possibilities may explain more false mem-
conflict condition (MChange in false memories = .10, SD = .15) ories in the salient intragroup conflict condition versus
as compared with the intergroup conflict condition both the in-group victim ambiguous conflict condition
(MChange in false memories = .04, SD = .13; t(39) = 1.91, and the intergroup conflict condition. Salient
p = .06; 95% CI [-.12, .003]; see Figure 5). intragroup conflict may divert eyewitnesses’ attention

2
An alternative explanation for the behavioral results shown thus far is that witnesses have more false memories when the
eyewitness and the perpetrator are in-group members. If the in-group perpetrator were the source of the memory errors, then
we would expect more memory errors in this condition. Alternatively, if the intragroup conflict created by seeing both the
victim and the perpetrator as members of one’s in-group engaged in conflict drove the increase in false memories, then we
would expect to see no difference in false memories between the in-group perpetrator-only and the in-group victim-only
conditions.
In order to test this, we recruited 28 young adults (Mage = 21.75 years, SD = 2.49, 17 female) from Indiana University to
complete the same task described in Experiment 1a and 1b with the only difference being that only the perpetrator or only the
victim’s group membership was identified. Thus, each participant viewed four slideshows that depicted an in-group perpetrator
with a non-identified victim (in-group perpetrator-only), and four slideshows that depicted an in-group victim with a non-
identified perpetrator (in-group victim-only). These slideshows were presented in pseudorandom order across participants.
True memories were calculated in the same manner as described in Experiments 1a and 1b. Here we found that there were
significantly more correct memories for the perpetrator only (MTrue memories = 15.64, SD = 6.30) as compared to the victim only
condition (MTrue memories = 13.79, SD = 6.05; t(27) = 2.43, p < .02; 95% CI [.29, 3.43]).
False memories were calculated as described in Experiments 1a and 1b. There was no significant difference in the false
memories for the perpetrator-only (MFalse memories = 7.39, SD = 5.68) and victim-only (MFalse memories = 7.25, SD = 4.82) conditions
(t(27) < 1, p = .82; 95% CI [-1.17, 1.45]).
These findings suggest that identifying the perpetrator as an in-group member did not affect the number of false memories.
Rather, the increase in the number of false memories observed in Experiments 1a and 1b was specific to the combination of
both the victim and the perpetrator being identified as in-group members.”
82 A. C. CARPENTER AND A. C. KRENDL

to the two group members, to either understand the Experiment 2

discrepancy between observed actions and the
To determine whether salient intragroup conflict
expected actions of the two group members, or
increased eyewitnesses’ false memories for crimes by
because the conflict itself was highly emotional. In
increasing the emotional or social salience of the crime
either case, participants’ attention may be diverted
we designed and implemented an fMRI study. If conflict
away from the details of the scene, resulting in weaker
elicited increased emotional salience, perceivers’ atten-
memory traces for these details, and greater suscept-
tion may have been drawn to the emotional element of
ibility to false memories.3 Consistent with the latter
a scene (the conflict), and away from peripheral, none-
interpretation, people show weaker memory for neu-
motional details (the details of the crime), thereby
tral peripheral elements of a scene given a highly
resulting in a memory trade-off (for review, see
emotional central event (Kensinger, Piguet, Krendl, &
Kensinger, 2009). Indeed, eyewitnesses have more
Corkin, 2005).
powerful affective responses to crimes committed by
A limitation to using a behavioral approach to under-
in-group versus out-group perpetrators (Marcus-
stand why salient intragroup conflict (relative to in-group
Newhall et al., 2002). However, the intragroup conflict
victim ambiguous conflict) elicited more false memories
may increase false memories by augmenting the con-
is that both possibilities (e.g., heightened social salience
flict’s social salience. That is, participants may have
related to the crime or heightened emotional salience)
focused on trying to understand why two in-group
predict the same behavioral prediction: increased false
members were in conflict. Supporting this idea, people
memories. Because the neural mechanisms underlying
have worse memory for negative behaviors about in-
social versus emotional salience have been well-charac-
group versus out-group members, suggesting a weaker
terized (e.g., Adolphs, 2009; Kensinger, 2009), we next
memory trace for expectation-inconsistent behaviors
leveraged neuroimaging to investigate why salient
(Howard & Rothbart, 1980).
intragroup conflict increased false memories.
Our neuroimaging experiment sought to disentangle
Several neuroimaging studies have evaluated false
the roles of emotional and social salience on false
memories (for review, see Schacter & Slotnick, 2004) by
memories for crimes involving salient intragroup con-
focusing on differences during the retrieval of true and
flict. A benefit to utilizing neuroimaging here is that the
false memories (Cabeza, Rao, Wagner, Mayer, &
neural correlates of emotional and social salience are
Schacter, 2001; Okado & Stark, 2003; Schacter,
each comprised of distinct networks (e.g., Adolphs,
Buckner, Koutstaal, Dale, & Rosen, 1997; Schacter
2009; Cloutier, Gabrieli, O’Young, & Ambady, 2011;
et al., 1996), rather than the contribution of mechan-
Kensinger, 2009; Murty, Ritchey, Adcock, & LaBar,
isms during encoding to subsequent false memories
2010). Specifically, emotionally salient information
(but, see Gonsalves et al., 2004; Okado & Stark, 2005;
engages the amygdala, fusiform gyrus, inferior frontal
Paller & Gonsalves, 2000). As a result, it remains
gyrus, and lingual gyrus (Adolphs, Denburg, & Tranel,
unknown whether increased activation in a specific set
2001; Adolps, Tranel, & Buchanan, 2005; Kensinger &
of neural regions during encoding reliably predicts false
Schacter, 2006; Murty et al., 2010; Waring & Kensinger,
memories. However, the dissociable neural regions
2011), while the anterior cingulate cortex (ACC), dor-
engaged in response to social as compared to emotion-
somedial prefrontal cortex (dmPFC), precuneus, inferior
ally salient information have been well-identified. We
frontal gyrus, and bilateral temporoparietal junction
therefore had two goals in Experiment 2: first, to deter-
interact when evaluating social information (e.g.,
mine whether intragroup conflict increased social or
Cloutier et al., 2011).
emotional salience during encoding; and second, to
If emotional salience disrupts memory given
identify the neural regions engaged during encoding
intragroup conflict, we expected heightened activation
that predicted subsequent false memory.
in the emotional salience network (e.g., amygdala,

3
The important distinction between a true and false memory is the incorporation of post-event misinformation in the memory for
the original event. While remembering contradictory information presented in the audio-recording as coming from the
slideshow results in a false memory, we are not arguing that the original memory trace is replaced or forgotten. Rather,
when the original memory trace is weakened, participants may be more likely to incorporate post-event misinformation (please
see the description on page 4 of the manuscript). However, if a discrepancy is detected or the post-event misinformation is not
incorporated into the memory of the original event, we argue that the original memory trace may still be retrieved. However,
there are many other factors after the encoding phase that may affect whether or not the original trace will be successfully
retrieved. Thus, our results are only able to comment on the effect of social salience during encoding of the original event;
however, we acknowledge that future work should evaluate the continuous interaction between encoding and retrieval in
relation to the current results.
 SOCIAL NEUROSCIENCE 83

fusiform gyrus, inferior frontal gyrus, and lingual gyrus) of Cognitive Neurology, London, UK). Data underwent
when individuals viewed crimes involving in-group standard preprocessing to remove sources of noise and
members. If, however, this conflict disrupts memory artifact. Preprocessing and analyses of functional data
via social salience, we predicted heightened activation were conducted in SPM8 (Wellcome Trust Centre for
in neural regions associated with cognitive conflict, Neuroimaging, London, UK). Images were realigned to
such as the ACC (Kerns et al., 2004). correct for motion, normalized to the Montreal
Neurological Institute (MNI) template, and spatially
smoothed (8 mm full-width-at-half-maximum [FWHM])
Materials and methods
using a Gaussian kernel. Using custom artifact detection
Participants software to detect motion artifact (http://www.nitrc.org/
projects/artifact_detect), individual runs were analyzed on
In order to parallel the sample size in Experiment 1b, we
a participant-by-participant basis to find outlier time
recruited 39 right-handed young adults without neurolo-
points. Specifically, we excluded volumes during which
gical or psychiatric diagnosis (Mage = 21.6 years,
participant head motion exceeded 1 mm and volumes in
SD = 2.2 years, 26 female) from Indiana University who
which the overall signal for that time point fell three SDs
participated for monetary compensation. One participant
outside the mean global signal for the entire run. Outlier
was excluded from analyses due to excessive movement.
time points were excluded from the GLM analysis via the
The Indiana University IRB approved this study.
use of participant-specific regressors of no interest.
Analyses were conducted on two levels: GLM and
Materials ROI analyses. GLM analyses incorporated separate task-
specific regressors for critical information trials that
The methods and materials are the same as described
later became a true memory, false memory, neither a
in Experiment 1a, with three exceptions: first, as in
true nor a false memory, and all noncritical information,
Experiment 1b, we only presented participants with
along with covariates of no interest (a session mean, a
the six crimes, each with 50 events (resulting in 300
linear trend, and six movement parameters derived
events per participant); second, conflict condition was
from realignment corrections). These regressors were
manipulated between subject; and third, encoding of
used to compute parameter estimates (β) and t-contrast
the original event took place while participants simul-
images (containing weighted parameter estimates) for
taneously underwent fMRI. The second session (intro-
each comparison at each voxel and for each subject.
duction of misinformation and memory task) was
Region of interest (ROI) analyses were conducted using
conducted outside the scanner.
the functional ROIs tool in SPM8 (marsbar; http://mars
bar.sourceforge.net/). Each ROI was extracted by using
fMRI procedure the contrast from each condition relative to baseline
fixation to conduct a region of interest (ROI) analysis.
Anatomical and functional whole-brain imaging was
Functional connectivity was examined using the
performed on a 3.0 T Siemens Trio Scanner (Trio,
generalized psychophysiological interactions (gPPI:
Siemens Ltd., Enlargen, Germany) using standard data
http://brainmap.wisc.edu/PPI; McLaren, Ries, Xu, &
acquisition protocols. Anatomical images were acquired
Johnson, 2012) toolbox in SPM8. The gPPI toolbox
using a high-resolution 3-D magnetization prepared
automatically accommodates multiple task conditions
rapid gradient echo sequence (MP-RAGE; 160 sagittal
in the same PPI model and compares functional con-
slices, TE = 2.67 ms, TR = 2000 ms, flip angle = 9°,
nectivity with a single seed region across conditions.
1 × 1 × 1 mm voxels). Functional images were collected
Each seed region was used to create volumes of interest
in six functional runs of 163 time points each, using a
(VOIs) for each subject by creating a 6 mm sphere
fast field echo-planar sequence sensitive to blood-oxy-
around this peak coordinate. Within each subject, the
gen level-dependent contrast (T2*) (35 axial slices per
gPPI toolbox was used to estimate functional connec-
whole-brain volume, 3 mm in-plane resolution, 3.8 mm
tivity across the entire brain with the seed region in the
thickness, 0 mm skip, TR = 2000 ms).
four memory type conditions (i.e., true memory, false
memory, generic, or guess) and to calculate 2 contrasts
Data analysis of interest (True memory > False memory and False
memory > True memory). The generic memory condi-
fMRI analyses
tion was used to define trials for which no correspond-
fMRI data were analyzed using the general linear model ing information was presented in the narrative (i.e.,
for event-related designs in SPM8 (Wellcome Department could not result in a false memory).
84 A. C. CARPENTER AND A. C. KRENDL

At the group level, contrasts from each of the indi- ambiguous conflict conditions. Replicating Experiment
vidual–subject gPPIs were entered into separate single- 1a, there were more false memories in the intragroup
sample t-tests. The first set of contrasts identified conflict condition (MFalse Memory = 14.19, SD = 6.40) than
regions more positively correlated with the seed region the ambiguous conflict condition (MFalse Memory = 8.55,
for true memories as compared to false memories. The SD = 4.39; t(36) = 3.23, p = .003).
second set of contrasts identified regions more posi- We conducted 2 (conflict type: ambiguous or sali-
tively correlated with the seed region for false mem- ent) × 2 (time: pre-crime realization or post-crime reali-
ories as compared to true memories. zation) mixed ANOVA with proportion false memories
We had two main goals in the fMRI analyses. First, we as the dependent variable. As with Experiments 1a and
examined how conflict condition affected the neural 1b, we used proportion false memories here because
activity underlying subsequent true and false memories. the number of possible items for which participants
Second, we examined how patterns of neural activity could have had false memories differed pre- and post-
differed by conflict condition (salient intragroup conflict crime realization. Replicating Experiment 1a, there was
versus in-group-victim ambiguous conflict) and by time a main effect of condition, F(1,36) = 8.64, p = .006,
(pre-crime realization versus post-crime realization), irre- η2partial = .19, because there more false memories in
spective of subsequent memory. We were unable to the intragroup conflict condition versus the ambiguous
consider subsequent memory for the latter analyses conflict condition (MIntragroup = .20, SD = .09;
because many participants did not have a sufficient MAmbiguous = .12, SD = .06; t(36) = 3.23, p = .003).
number of false memories in just the pre- or post- There was a main effect time, F(1,36) = 13.09,
crime realization conditions to detect reliable differences. p = .001, η2partial = .27, and a trending interaction
With respect to the subsequent memory analysis for between time and conflict condition (salient intragroup
our first goal, we excluded five participants because or ambiguous), F(1,36) = 2.81, p = .10, η2partial = .07.
they had fewer than five true memories, and six others Although this interaction was marginal, the directional-
because they had fewer than five false memories (see ity of the effects and the overall effect size replicated
Okado & Stark, 2005, for similar exclusionary criteria). the pattern of results from both Experiment 1a and
This left 27 remaining participants (Mage = 21.4 years, Experiment 1b. Critically, replicating our results in
SD = 2.1 years, 19 female), 14 of whom had been Experiment 1a, there were more false memories over-
pseudorandomly assigned to the intragroup conflict time in the salient intragroup as compared to the
condition, and 13 of whom were assigned to the ambiguous group conditions (MIntragroup = .23,
ambiguous conflict condition. Given the relatively SD = .11; MAmbiguous = .13, SD = .09; t(36) = 3.01,
small sample size in this condition, data were extracted p = .005).
at a threshold of p < .005 using a more conservative 15- True memories. We conducted a separate t-test to
voxel extent threshold (see Woo, Krishnan, & Wager, compare the overall number of true memories in the
2014 for discussion on cluster thresholding). salient and ambiguous conflict conditions. Like
Importantly, subsequent memory performance did Experiment 1a, there was no difference in the number
not place any constraints on our sample size with of true memories for the intragroup conflict (MTrue
respect to our second goal of examining the effect of Memory = 22.19, SD = 11.32) as compared to the ambig-
condition pre- and post- crime realization. Thus, these uous conflict conditions (MTrue Memory = 19.64,
results are reported with the full sample of 38 partici- SD = 11.91; t(36) < 1, p = .51).
pants (16 = intragroup conflict condition, 22 = ambigu-
ous conflict condition). Thus, an additional benefit of
fMRI results
this analysis was that it allowed us to validate the
results from the smaller subsequent memory sample
with a sample that was more than one-third larger at
Effects of conflict condition on subsequent memory
a more conservative threshold of p < .001 uncorrected,
with 10 contiguous voxels. We first examined how conflict condition affected
neural activity underlying subsequent true and false
memories using a 2 (conflict condition: intragroup or
Results ambiguous) X 2 (memory type: true or false) whole-
brain ANOVA using the respective task > baseline con-
Behavioral results
trasts for each variable. Results revealed a main effect of
False memories. We first conducted a t-test to compare memory type in the left superior frontal gyrus (BA 6;
the overall number of false memories in the salient and Table 3), such that this region was more active when
 SOCIAL NEUROSCIENCE 85

Table 3. Main effects of memory type, main effect of conflict condition, and memory type × conflict condition
interaction that emerged from the 2 (conflict condition: intragroup or ambiguous) × 2 (memory type: true
memory or false memory) whole-brain voxelwise ANOVA.
Brain Region x y z Kextent T-score
Main effect of memory
True Memory > False Memory
Left superior frontal gyrus (BA 6) −21 6 72 27 3.92
Left anterior cingulate cortex (BA 32) −21 12 39 23 2.60
False Memory > True Memory
No significant voxels
Main effect of conflict condition
Intragroup conflict > Ambiguous conflict
Left superior frontal gyrus (BA 10) −24 63 12 34 3.14
Right superior frontal gyrus (BA 8/10) 3 39 57 129 4.05
Right superior frontal gyrus (BA 6/8) 27 30 57 64 3.35
Right middle frontal gyrus (BA 10) 42 57 6 203 4.47
Right dmPFC (BA 9) 18 57 42 27 3.53
Right dlPFC (BA 9/46) 42 45 33 173 5.01
Left dlPFC (BA 46) −48 39 33 56 3.76
Right orbitofrontal cortex (BA 11) 15 57 −15 18 3.2
Left orbitofrontal cortex (BA 11) −24 45 −18 94 4
Right middle frontal gyrus (BA 11) 36 36 −21 49 3.62
Left inferior frontal gyrus (BA 47) −18 36 −6 * 3.22
Right inferior frontal gyrus (BA 47) 27 18 −12 * 4.07
Left Aanterior cingulate (BA 32) −12 30 −6 24 3.24
Right superior temporal gyrus (BA 38) 18 15 −33 181 5.57
Left parahippocampal gyrus (hippocampus) −24 −12 −21 * 3.16
Right parahippocampal gyrus (BA 19/30) 24 −45 0 97 3.58
Left sub-gyral (BA 20) −39 −15 −21 33 3.33
Left parahippocampal gyrus (BA 30) −6 −36 6 * 3.15
Right middle temporal gyrus (BA 39) 51 −72 15 53 3.4
Left middle temporal gyrus (BA 39) −45 −75 27 * 3.08
Left inferior parietal lobule (BA 40) −30 −39 39 74 3.52
Right inferior parietal lobule (BA 40) 45 −45 60 1797 5.82
Left cingulate gyrus (BA 31) −21 −45 21 96 3.61
Left fusiform gyrus (BA 37) −60 −48 −18 91 4.08
Right cuneus (BA 17) 9 −78 12 16 3.21
Right cuneus (BA 19) 18 −93 42 356 3.82
Left middle occipital gyrus (BA 19) −39 −84 21 56 3.99
Left lingual gyrus (BA 18) −15 −102 −6 33 3.09
Left cerebellum 0 −63 −30 24 3.42
Left cerebellum −33 −72 −42 2450 5.99
Ambiguous conflict > Intragroup conflict
No significant voxels
Memory type X conflict condition
Ambiguous conflict True memory> False memory
Right inferior parietal lobule (BA 40) 54 −30 30 61 2.97
Right postcentral gyrus (BA 3) 33 −36 48 25 3.4
Intragroup conflict True memory> False memory
Left superior temporal gyrus (BA 38) −24 24 −33 21 3.32
Subcluster activations noted with *. Organized by region anterior to posterior. Coordinates are MNI.

individuals evaluated images that were subsequently conflict conditions. No regions were more active toward
remembered. No region was more active for items ambiguous conflict versus intragroup conflict condi-
that were subsequently falsely remembered. tions. The only peak emerging from the memory type
A main effect of conflict condition elicited a wide × conflict condition interaction was within left postcen-
range of neural activity, including bilateral dorsolateral tral gyrus (BA 1/2; see Table 3).
prefrontal cortex (BA 46), right middle and superior
frontal gyrus (BA 6/8/11), right inferior frontal gyrus
(BA 47), right middle and superior temporal gyrus (38/
Effects of crime realization and conflict condition
39), right parahippocampal gyrus, bilateral inferior par-
on neural activity
ietal lobule (BA 40), and visual processing regions To identify neural activity differing pre- and post-
including fusiform gyrus and right cuneus (see Table 3 crime realization in the intragroup conflict versus
for complete list of activations). These regions were the ambiguous conflict conditions, we conducted a
more active in the intragroup conflict versus ambiguous 2 (conflict condition: intragroup or ambiguous) × 2
86 A. C. CARPENTER AND A. C. KRENDL

(time: pre-crime realization v. post-crime realization) investigated which neural regions were functionally
whole-brain ANOVA using all trials in which partici- coupled with the ACC for items that were subsequently
pants viewed critical information (i.e., specific items falsely remembered (using gPPI). We conducted this
for which they later received misinformation). A main analysis for both conflict conditions by creating a seed
effect of conflict condition elicited activity in bilateral region for the right ACC peak emerging from the t-test
precuneus, ACC, dmPFC, and fusiform gyrus (see contrasting the two conflict conditions post-crime rea-
Table 4 for complete list of activations). These heigh- lization (3, 33, 24; see Table 5). We then identified
tened activations were driven by activity toward neural regions that were functionally coupled with
intragroup versus ambiguous conflict. No regions those regions on items for which participants in the
were more active for the ambiguous versus intragroup conflict condition had a subsequent false
intragroup conflict conditions. A main effect of time (as compared to true) memory. We conducted the
emerged in regions broadly associated with social same analysis for participants in the ambiguous conflict
cognition, including dmPFC, orbitofrontal cortex, condition. For the intragroup conflict condition, we
and right temporoparietal junction. These activations found heightened functional connectivity between the
were driven by the post-crime versus pre-crime reali- right ACC and the right dmPFC (see Table 6). For the
zation. No regions were more active pre-crime reali- ambiguous conflict condition, however, the rACC was
zation as compared to post (see Table 4 for complete functionally coupled with the posterior cingulate, and
list of activations). No regions emerged in the conflict not the dmPFC. See Figure 7.
condition × time interaction.
Discussion
T-test comparing intragroup and ambiguous
Results revealed evidence of neural activation in
conflict post-crime realization
regions associated with processing social salience (i.e.,
Because behavioral analyses revealed a significant inter- ACC, dmPFC), which predicted participants’ subsequent
action for false memories (driven by a greater increase false memories. Although these results emerged from
in false memories from pre-crime realization to post- analyses on a smaller subset of our data (because not
crime realization for the intragroup conflict condition as all participants had a sufficient number of false mem-
compared with the ambiguous conflict condition), we ories), the same pattern of results emerged when we
conducted a t−test to directly compare the neural acti- compared neural activation post- as compared to pre-
vation in the two conflict conditions during post-crime crime realization on the full dataset. Specifically, we
realization. Post-crime realization, participants in the found heightened activation in regions associated
intragroup conflict condition (as compared to the with social salience (including ACC) post-crime realiza-
ambiguous conflict condition) had greater activation tion in the intragroup conflict condition. Functional
in neural regions associated with social cognition, connectivity analyses demonstrated that the ACC was
including in the right ACC, bilateral inferior frontal functionally coupled with dmPFC. Thus, both analyses
gyrus, bilateral fusiform gyrus, and bilateral precuneus provide converging evidence that false memories from
(see Figure 6; and Table 5 for complete list of activa- intragroup conflict emerge because the conflict elicits
tions). No regions were more active post-crime realiza- increased social, not emotional, salience.
tion in the ambiguous conflict condition as compared
to the intragroup conflict condition.
General discussion
Results of these experiments suggest that individuals
Regions exhibiting differential neural connectivity
are more susceptible to false memories after witnessing
with the anterior cingulate cortex as a function of
crimes depicting salient intragroup conflict versus sali-
subsequent false memories
ent out−group conflict, or mixed-group conflict (with
The above results demonstrated that participants in the an in−group or out−group victim). Our neuroimaging
intragroup conflict condition (as compared to the findings suggest that the social salience of the
ambiguous conflict condition) had greater activation intragroup conflict may be associated with increased
post-crime realization in neural regions associated susceptibility to false memories. Specifically, we found
with intragroup conflict, including in the right ACC heightened activation in neural regions associated with
and bilateral inferior frontal gyrus (Amodio, 2014). social salience (e.g., ACC and dmPFC), but not emo-
Because prior work has widely implicated the ACC in tional salience, in the salient intragroup conflict condi-
conflict resolution (e.g., Kerns et al., 2004), we tion. Further, activity in these regions was more
 SOCIAL NEUROSCIENCE 87

Table 4. Main effects of conflict condition, time, and conflict condition × time interaction that emerged from
the 2 (conflict condition: intragroup or ambiguous) × 2 (time: pre-crime realization or post-crime realization)
whole-brain voxelwise ANOVA.
Brain Region x y z Kextent T-score
Main effect of conflict condition
Intragroup > Ambiguous
Left dmPFC (BA 9/10) −3 63 24 115 4.5
Right dlPFC (BA 10) 33 63 3 51 4.4
Right superior frontal gyrus (BA 10) 21 57 −9 12 3.92
Left medial frontal gyrus (BA 8) −3 54 42 40 3.87
Right superior frontal gyrus (BA 8) 45 18 51 44 4.08
Right superior frontal gyrus (BA 8) 30 36 51 24 3.74
Right superior frontal gyrus (BA 9/10) 42 42 30 150 5.23
Left superior frontal gyrus (BA 6) 0 36 60 38 4.18
Left middle frontal gyrus (BA 8) −42 30 39 24 3.73
Left middle frontal gyrus (BA 6) −36 −3 48 29 3.87
Right middle frontal gyrus (BA 6) 18 −12 57 6499 5.71
Right fusiform gyrus (BA 20) 57 −33 −21 * 5.51
Left anterior cingulate gyrus (BA 32) −3 21 42 231 4.32
Left middle temporal gyrus (BA 38) −45 9 −42 17 3.65
Right transverse temporal gyrus (BA 41) 33 −39 12 23 4.02
Right middle temporal gyrus (BA 19) 51 −75 18 11 3.57
Left parahippocampal gyrus (BA 36) −24 −21 −30 10 3.66
Right inferior parietal lobule (BA 40) 39 −30 39 83 4.28
Right inferior parietal lobule (BA 7/40) 45 −54 60 327 5.36
Left inferior parietal lobule (BA 40) −33 −54 36 36 3.81
Left superior parietal lobule (BA 7) −42 −57 60 29 3.67
Left precuneus (BA 7) −6 −78 39 440 4.84
Right precuneus (BA 7) 18 −81 51 15 3.43
Left cerebellum 0 −60 −42 41 3.87
Ambiguous > Intragroup
No significant voxels
Main effect of time
Post-crime realization > Pre-crime realization
Left superior frontal gyrus (BA 10) −33 60 −3 56 4.41
Right anterior cingulate (BA 32) 3 33 24 160 4.62
Left middle frontal gyrus (BA 46) −48 39 33 24 4.15
Right inferior frontal gyrus (BA 45/47) 60 15 0 112 5.35
Right inferior frontal gyrus (BA 47) 39 15 −6 127 4.69
Right precentral gyrus (BA 6) 30 −12 57 77 4.91
Left insula (BA 13) −39 18 3 195 5.48
Right postcentral gyrus (BA 2) 30 −24 36 52 4.72
Left postcentral gyrus (BA 3) −27 −30 60 30 4.14
Left inferior parietal lobule (BA 40) −51 −48 48 18 3.94
Right superior parietal lobule (BA 7) 45 −57 60 86 5.38
Left fusiform gyrus (BA 20) −48 −30 −30 75 4.44
Right fusiform gyrus (BA 20) 42 −21 −24 97 5.43
Right uncus (BA 36) 24 −12 −39 * 4.92
Right parahippocampal gyrus (BA 36) 27 −18 −27 * 3.72
Left uncus (BA 36) −21 −9 −39 142 4.73
Left parahippocampal gyrus (BA 36) −27 −18 −24 * 4.67
Left superior temporal gyrus (BA 42) −69 −30 12 25 4.2
Left middle temporal gyrus (BA 21) −69 −33 −6 * 3.88
Right inferior temporal gyrus (BA 20/21) 60 −33 −21 225 6.91
Right inferior temporal gyrus (BA 20) 57 −60 −15 22 3.99
Left precuneus (BA 7/24) 0 −36 45 275 4.88
Left precuneus (BA 7) −6 −78 39 33 4.12
Right precuneus (BA 7) 9 −69 42 57 4.35
Right lingual gyrus (BA 17) 21 −102 −9 25 3.92
Right thalamus 3 −12 3 18 4.27
Right lentiform nucleus 15 −9 −3 22 3.86
Right cerebellum 33 −42 −39 60 4.64
Left cerebellum −42 −51 −39 49 4.4
Right cerebellum 45 −75 −42 52 4.09
Ambiguous conflict > Intragroup conflict
No significant voxels
Memory type × conflict condition
No significant voxels
Subcluster activations noted with *. Organized by region anterior to posterior. Coordinates are MNI.
88 A. C. CARPENTER AND A. C. KRENDL

an event affects participants’ susceptibility to source

misattributions. The current study suggests crimes
involving a victim and perpetrator who have the same
group membership to each other, and (critically) to the
eyewitness, may elicit more false memories.
Second, our results demonstrated that false memories
overall increased from pre-crime realization to post-
crime realization. That is, even when controlling for dif-
ferences in the number of memories assessed from pre-
crime realization to post-crime realization (by using a
proportion score), our results across Experiments 1a, 1b,
and 2 demonstrated that participants overall had more
false memories in post-crime realization than pre-crime
realization, and this effect was specifically exacerbated
Figure 6. T-test results showing increased activation in the for crimes depicting intragroup conflict. The present
rACC during the critical trials in intragroup conflict as compared research therefore provided support for the role of con-
to ambiguous conflict conditions post-crime realization.
flict resolution in memory for events specifically invol-
ving intragroup conflict.
pronounced after the nature of the conflict became The neuroimaging results in Experiment 2 further
clear (e.g., post-crime realization). elucidated the mechanisms underlying our findings.
The current study extends the literature on false Participants in the intragroup conflict condition (as
memories in several important ways. Of central impor- compared to the ambiguous conflict condition) had
tance, prior research has not evaluated the role in heightened activation post-crime realization in neural
which the identification of specific individuals within regions associated with social cognition, notably those

Table 5. Regions that were more active as a function of conflict condition post-crime realization.
Brain Region x y z Kextent T-score
Ambiguous conflict > Intragroup conflict
No significant voxels
Intragroup conflict > Ambiguous conflict
Left superior frontal gyrus (BA 10) −33 60 −3 56 4.41
Left dlPFC (BA 46) −48 39 33 24 4.15
Left inferior frontal gyrus (BA 46) −54 36 6 14 4.04
Right ACC/dmPFC (BA 32) 3 33 24 160 4.62
Left insula (BA 13) −39 18 3 195 5.48
Right inferior frontal gyrus (BA 45) 60 15 0 112 5.35
Right inferior frontal gyrus (BA 47) 39 15 −6 127 4.69
Left insula (BA 13) −30 −33 24 14 3.88
Right lentiform nucleus 15 −9 −3 22 3.86
Right precentral gyrus (BA 6) 30 −12 57 77 4.91
Right thalamus 3 −12 3 18 4.27
Right parahippocampal gyrus (BA 36) 27 −18 −27 * 3.72
Left parahippocampal gyrus (BA 36) −27 −18 −24 142 4.67
Right fusiform gyrus (BA 20) 42 −21 −24 97 5.43
Left fusiform gyrus (BA 20) −48 −30 −30 75 4.44
Right postcentral gyrus (BA 2) 30 −24 36 52 4.72
Left postcentral gyrus (BA 3) −27 −30 60 30 4.14
Left superior temporal gyrus (BA 21/42) −69 −30 12 25 4.2
Right inferior temporal gyrus (BA 20/21) 60 −33 −21 225 6.91
Right inferior temporal gyrus (BA 20) 57 −60 −15 22 3.99
Left inferior parietal lobule (BA 40) −51 −48 48 18 3.94
Right superior parietal lobule (BA 7) 45 −57 60 86 5.38
Left lingual gyrus (BA 18) −12 −51 3 14 3.8
Right lingual gyrus (BA 17) 21 −102 −9 25 3.92
Left precuneus (BA 7) 0 −36 45 275 4.88
Left precuneus (BA 7) −6 −78 39 33 4.12
Right precuneus (BA 7) 9 −69 42 57 4.35
Right cerebellum 33 −42 −39 60 4.64
Right cerebellum 45 −75 −42 52 4.09
Right cerebellum 15 −78 −33 11 3.6
Left cerebellum −42 −51 −39 49 4.4
Subcluster activations noted with *. Organized by region anterior to posterior. Coordinates are MNI.
 SOCIAL NEUROSCIENCE 89

Table 6. Regions that were functionally coupled with the right may disrupt their ability to accurately encode the
ACC for participants in the intragroup conflict condition when details of the crime, thereby increasing their suscept-
they viewed critical items for which they subsequently had a ibility to false memories. Consistent with this specula-
false (as compared to true) memory. Coordinates are MNI.
tion, participants in the intragroup conflict condition
Brain Region x y z Kextent T-score
also experienced more false memories post-crime reali-
Ambiguous conflict condition
Left posterior cingulate (BA 31) −18 −54 24 16 4.73 zation as compared to participants in the ambiguous
Left dlPFC (BA 10) −30 42 21 15 4.49 conflict condition.
Intragroup conflict condition Our results are consistent with previous research on
Left middle frontal gyrus (BA 6) −15 9 51 59 4.58
Right dmPFC 18 18 51 56 4.08 false memory (e.g., Cabeza et al., 2001; Kim & Cabeza,
2007; Okado & Stark, 2003), demonstrating the role of
the prefrontal cortex, temporal lobes, and parietal cortex
in encoding information that subsequently becomes a
true as compared to a false memory. Notably, this
research has shown that the prefrontal cortex, parietal
cortex, and late visual processing regions are engaged
when individuals encode information for which they
subsequently form a true or false memory (e.g., Kim &
Cabeza, 2007; Okado & Stark, 2003). However, the poster-
ior medial temporal lobes (e.g., notably the parahippo-
campal gyrus), and early visual areas are more active
when individuals encode information that later becomes
a true memory (Cabeza et al., 2001; Kim & Cabeza, 2007;
Okado & Stark, 2003).
At first glance, our finding that intragroup conflict
increased false memories for details related to the crime
Figure 7. Activity in the dmPFC was functionally coupled with by increasing the social salience of the conflict appears
the right ACC when participants in the intragroup conflict to be consistent with extensive research on the mem-
condition viewed items for which they later had false (as ory trade-off effect (for review, see Hamann, 2001).
compared to true) memories.
Simply put, when participants evaluate emotional
scenes, they have enhanced memories for central
involved with resolving conflicts. Specifically, compar- item-specific details about a negative event or scene,
ing neural activation in the two conflict conditions post- but their memory accuracy for neutral, peripheral
crime realization revealed heightened activation in ACC details is disrupted (for review, see Kensinger, 2009).
—a region that has been widely implicated in conflict However, an important distinction in the current inves-
monitoring and the engagement of cognitive control tigation is that although emotionally salient information
(e.g., Kerns et al., 2004)—and the inferior frontal gyrus may enhance memory for specific details, social infor-
—a region broadly implicated in initiating cognitive mation (e.g., impression incongruities) may lead to
control in social judgments (e.g., Amodio, 2014). The meta-perceptions about the conflict itself and result in
ACC has also been implicated as being more engaged similar heightened false memory rates. Importantly,
when participants view social targets that they perceive social salience would only affect participants’ memory
to be more similar to themselves (Leshikar, Cassidy, & for items presented once the conflict becomes appar-
Gutchess, in press). Moreover, a functional connectivity ent (i.e., post-crime realization). However, there are
analysis revealed that heightened activation in the ACC many other factors after the encoding phase that may
was associated with heightened functional coupling in affect whether or not the original trace will be success-
the dmPFC when participants in the intragroup conflict fully retrieved. Thus, our results are only able to com-
condition evaluated items for which they subsequently ment on the effect of social salience during encoding of
had false memories. This coupling was not present in the original event. Future work should evaluate the
the ambiguous conflict condition. Thus, one possible, continuous interaction between encoding and retrieval
albeit speculative, explanation for this pattern of results in relation to the current results.
is that during intragroup conflict, participants may be Our finding that intragroup conflict increased the
trying to resolve the discrepancy between their expec- social salience of the crime is consistent with previous
tation of how in−group members should act and the research on the Black Sheep Effect, which suggests that
witnessed actions of said in−group members, which perceivers feel less affiliation with in−group members
90 A. C. CARPENTER AND A. C. KRENDL

who violate social norms, but stronger affiliation with in 2001). In the current study, the victims, perpetrators,
−group victims of social norm violations (Eidelman & and participants were Caucasian. Future research
Biernat, 2003; Marcus-Newhall et al., 2002; Marques & should nonetheless examine this alternate interpreta-
Yzerbyt, 1988; Nourkova et al., 2004; Pinto et al., 2010). tion. It is important to note that the misinformation on
This polarized response may be more robust when an in which the false memories were based stemmed from
−group member is victimized by another in−group questions about the peripheral details of the crime, not
member (e.g., Eidelman & Biernat, 2003). That is, about the perpetrators or the victims themselves.
intragroup conflict may increase susceptibility to false Therefore, eyewitnesses may have had better memories
memories because perceivers may fixate on the social for the central details of the crimes (e.g., the perpetrator
aspect of the conflict rather than distancing themselves or the victim), and their increased false memories for
from the in−group perpetrator (e.g., resolving the the peripheral details may result from a memory trade
expectation-based incongruity associated with seeing −off. However, because we did not manipulate the
two in−group members in conflict). Consistent with central or peripheral details in the current study, this
this interpretation, recent research suggests that indivi- possibility is speculative. Future research should exam-
duals have better memories for impressions they form ine this question, with respect to whether memory for
of people they perceive to be similar to themselves, but the central details of the crimes (e.g., the in−group
only provided the impressions are positive (e.g., victim and in−group perpetrator) is enhanced. Further,
Leshikar, Dulas, & Duarte, 2015; Leshikar, Park, & future research should utilize eye-tracking in concert
Gutchess, 2015), which is likely not the case in the with fMRI methods in order to better understand the
intragroup conflict condition. role of attention allocation in response to both emo-
Although we can only speculate as to the reasons tional and social salience and how the potential pat-
why participants’ proportion of false memories terns of attention diversion affect subsequent false
increased post-crime realization, one intriguing possibi- memory for both central and peripheral details.
lity that should be examined in future research is that A related limitation to the current research is that
the increase in false memories reflects the effect of the stimuli we used did not allow us to disentangle
increased arousal or stress on memory. That is, partici- whether witnesses’ increased susceptibility to false
pants might have found the event information to be memories post-crime (as compared to pre-crime reali-
more stressful and/or arousing when it occurred post- zation) affected their overall ability to accurately iden-
crime realization because that is when they understood tify the perpetrator. In the current study, we were
that they were viewing a crime. In support of this constrained by the limitations of the stimuli used in
assertion, we found that events presented post-crime prior studies. Thus, future research may develop a
realization (regardless of conflict condition) were more more tightly controlled stimuli set that could address
susceptible to false memories than those presented the question of whether the overall gist of the witness’
pre-crime details. However, the fact that intragroup memory (e.g., based on the pre-crime realization
conflict exacerbated these effects suggests that the details) is sufficient to ensure that the perpetrator is
conflict might have introduced even more arousal properly identified.
and/or stress. Indeed, valence and arousal have both Second, our in−group/out−group manipulation
been widely shown to affect memory (e.g., Kensinger & focused solely on University affiliation, while in real-
Corkin, 2004), as has stress (see Vedhara, Hyde, Gilchrist, world scenarios there are likely multiple group mem-
Tytherleigh, & Plummer, 2000). Future research using bership features (e.g., gender, race, etc.) that affect
psychophysiological measures should disentangle these individuals’ affiliation with the in−group. Thus, future
possibilities. research should evaluate how the level of affiliation
An important limitation of this study is that we did within the in−group (i.e., university affiliation versus
not measure participants’ memory for faces. Therefore both university affiliation and gender affiliation) may
we cannot discount the alternate possible interpreta- affect subsequent memory.
tion that participants’ had more misinformation in the There are several limitations to the standard misinfor-
intragroup conflict condition because they allocated mation paradigm. First, each participant evaluated many
more attentional resources to in-group faces as com- different events, and several variables were manipulated
pared to out-group faces (e.g., Ito & Urland, 2005; within subject. It is possible that presenting several events
Trawalter, Todd, Baird, & Richeson, 2008). However, involving intragroup conflict between two in−group
prior research showing a memory enhancement for in members affected participants’ expectations for future
−group as compared to out−group faces has focused events. Namely, participants may have expected
primarily on race (Golby, Gabrieli, Chiao, & Eberhardt, intragroup conflict in all events as compared to an
 SOCIAL NEUROSCIENCE 91

unidentified individual or an out−group member in con- A. Carpenter drafted the manuscript, and A. Krendl provided
flict with a member of the in−group. However, the pre- critical revisions. Both authors approved the final version of
the manuscript for submission.
sentation of the slideshow events and the misinformation
audio−recordings was randomized, so any confounds
associated with general expectation violation would be
washed out. Additionally, participants were presented ORCID
with multiple events depicting criminal activity, while in Alexis C. Carpenter http://orcid.org/0000-0003-2198-3977
a real-world situation it seems highly unlikely that partici-
pants would witness six consecutive events depicting
conflict. However there are two features of our experi- References
mental design that speak against the possibility that
Abshire, J., & Bornstein, B. H. (2003). Juror sensitivity to the
exposure to multiple slideshows affected the reported
cross-race effect. Law and Human Behavior, 27(5), 471–480.
results. First, the randomization of both the order of the doi:10.1023/A:1025481905861
event slideshows and the misinformation narratives Adolphs, R. (2009). The social brain: Neural basis of social
would diminish the effects of fatigue or overexposure. knowledge. Annual Review of Psychology, 60, 693–716.
Second, the slideshows and audio-recordings have been doi:10.1146/annurev.psych.60.110707.163514
Adolphs, R., Denburg, N., & Tranel, D. (2001). The amygdala’s
reliably used to evaluate false memories both behaviorally
role in long-term declarative memory for gist and detail.
and using neuroimaging methods in several previous Behavioral Neuroscience, 115, 983–992. doi:10.1037/0735-
studies (Okado & Stark, 2005; Stark et al., 2010). 7044.115.5.983
Together, the results from the current study contribute Adolps, R., Tranel, D., & Buchanan, T. W. (2005). Amygdala
to the growing body of literature on false memories in damage impairs emotional memory for gist but not details
eyewitness testimony. Specifically, our results demon- of complex stimuli. Nature, 8(4), 512–518. doi:10.1038/nn1413
Amodio, D. M. (2014). The neuroscience of prejudice and
strated that the group membership of the victim and the
stereotyping. Nature Reviews Neuroscience, 15, 670–682.
perpetrator involved in a crime influences eyewitnesses’ doi:10.1038/nrn3800
susceptibility to false memories by increasing the social Bernstein, M., Young, S., & Hugenberg, K. (2007). The cross-
salience of the event during the encoding phase. While category effect: Mere social categorization is sufficient to elicit
the current results suggest that memories for details related an own-group bias in face recognition. Psychological Science,
18(8), 706–712. doi:10.1111/j.1467-9280.2007.01964.x
to crimes are malleable and can be influenced by group
Burt, C. D. B., Kemp, S., & Conway, M. (2004). Memory for true
membership during the encoding phase, future research and false autobiographical event descriptions. Memory, 12,
should investigate additional mechanisms that may play a 545–552. doi:10.1080/09658210344000071
role during the misinformation and retrieval phases. Cabeza, R., Rao, S. M., Wagner, A. D., Mayer, A. R., & Schacter,
D. L. (2001). Can medial temporal lobe regions distinguish
true from false? An event-related functional MRI study of
Acknowledgments veridical and illusory recognition memory. Proceedings of
the National Academy of Sciences of the United States of
The authors thank Eunice Lee for assistance with data collec- America, 98(8), 4805–4810. doi:10.1073/pnas.081082698
tion. Partial support for this research was provided by a summer Carpenter, A. C., & Schacter, D. L. (in press). Flexible retrieval:
research grant from the Hutton Honor’s College at Indiana When true inferences produce false memories. Journal of
University awarded to ACC. We thank Brittany Cassidy and Experimental Psychology: Learning, Memory & Cognition.
Helen Jing for their invaluable feedback on the manuscript. Cloutier, J., Gabrieli, J. D. E., O’Young, D., & Ambady, N. (2011).
An fMRI study of violations of social expectations: When
people are not who we expect them to be. NeuroImage, 57
Disclosure statement (2), 583–588. doi:10.1016/j.neuroimage.2011.04.051
No potential conflict of interest was reported by the authors. Devitt, A. L., Monk-Fromont, E., Schacter, D. L., & Addis, D. R.
(2016). Factors that influence the generation of autobiogra-
phical memory conjunction errors. Memory, 24(2), 204–222.
doi:10.1080/09658211.2014.998680
Funding Eidelman, S., & Biernat, M. (2003). Derogating black sheep:
This work was supported by the Hutton Honor’s College at Individual or group protection? Journal of Experimental
Indiana University. Social Psychology, 39(6), 602–609. doi:10.1016/S0022-1031
(03)00042-8
Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009).
Author contributions Statistical power analyses using G*Power 3.1: Tests for
correlation and regression analyses. Behavior Research
A. Carpenter and A. Krendl developed the study concept and Methods, 41, 1149–1160. doi:10.3758/BRM.41.4.1149
design. Testing and data collection were performed by A. Golby, A. J., Gabrieli, J. D. E., Chiao, J. Y., & Eberhardt, J. L.
Carpenter. A. Carpenter and A. Krendl conducted the analyses. (2001). Differential responses in the fusiform region to
92 A. C. CARPENTER AND A. C. KRENDL

same-race and other-race faces. Nature Neuroscience, 4(8), Lindsay, D. S., & Johnson, M. K. (1989). The eyewitness sug-
845–850. doi:10.1038/90565 gestibility effect and memory for source. Memory &
Gonsalves, B., Reber, P. J., Gitelman, D. R., Parrish, T. B., Mesulam, Cognition, 17, 349–358. doi:10.3758/BF03198473
M.-M., & Paller, K. A. (2004). Neural evidence that vivid imagin- Lindsay, R. C. L., Lim, R., Marando, L., & Cully, D. (1986). Mock-
ing can lead to false remembering. Psychological Science, 15 juror evaluations of eyewitness testimony: A test of meta-
(10), 655–660. doi:10.1111/j.0956-7976.2004.00736.x memory hypotheses. Journal of Applied Social Psychology,
Hamann, S. (2001). Cognitive and neural mechanisms of emo- 16(5), 447–459. doi:10.1111/j.1559-1816.1986.tb01151.x
tional memory. Trends in Cognitive Sciences, 5(9), 394–400. Loftus, E. F. (1975). Leading questions and the eyewitness
doi:10.1016/S1364-6613(00)01707-1 report. Cognitive Psychology, 7(4), 560–572. doi:10.1016/
Hardt, O., Nader, K., & Nadel, L. (2013). Decay happens: The 0010-0285(75)90023-7
role of active forgetting in memory. Trends in Cognitive Loftus, E. F. (1979). Eyewitness testimony. Cambridge, MA:
Sciences, 17(3), 111–120. doi:10.1016/j.tics.2013.01.001 Harvard University Press.
Howard, J. W., & Rothbart, M. (1980). Social categorization and Loftus, E. F. (2005). Planting misinformation in the human
memory for in-group and out-group behavior. Journal of mind: A 30-year investigation of the malleability of mem-
Personality and Social Psychology, 38(2), 301–310. ory. Learning & Memory, 12, 361–366. doi:10.1101/lm.94705
doi:10.1037/0022-3514.38.2.301 Loftus, E. F., Miller, D. G., & Burns, H. J. (1978). Semantic
Hupbach, A., Gomez, R., & Nadel, L. (2009). Episodic memory integration of verbal information into a visual memory.
reconsolidation: Updating or source confusion? Memory, 17 Journal of Experimental Psychology: Human Learning &
(5), 502–510. doi:10.1080/09658210902882399 Memory, 4, 19–31. doi:10.1037/0278-7393.4.1.19
Ito, T. A., & Urland, G. R. (2005). The influence of processing Marcus-Newhall, A., Blake, L. P., & Baumann, J. (2002).
objectives on the perception of faces: An ERP study of race Perceptions of hate crime perpetrators and victims as influ-
and gender perception. Cognitive, Affective, & Behavioral enced by race, political orientation, and peer group.
Neuroscience, 5(1), 21–36. doi:10.3758/CABN.5.1.21 American Behavioral Scientist, 46(1), 108–135. doi:10.1177/
Kensinger, E. A. (2009). Remembering the details: Effects of 0002764202046001008
emotion. Emotion Review, 1(2), 99–113. doi:10.1177/ Marques, J. M., & Yzerbyt, V. Y. (1988). The black sheep effect:
1754073908100432 Judgmental extremity towards ingroup members in inter-
Kensinger, E. A., & Corkin, S. (2004). Two routes to emotional and intra-group situations. European Journal of Social
memory: Distinct neural processes for valence and arousal. Psychology, 18(3), 287–292. doi:10.1002/ejsp.2420180308
Proceedings of the National Academy of Sciences of the McLaren, D. G., Ries, M. L., Xu, G., & Johnson, S. C. (2012). A
United States of America, 101(9), 3310–3315. doi:10.1073/ generalized form of context-dependent psychophysiologi-
pnas.0306408101 cal interactions (gPPI): A comparison to standard
Kensinger, E. A., Piguet, O., Krendl, A. C., & Corkin, S. (2005). approaches. NeuroImage, 61(4), 1277–1286. doi:10.1016/j.
Memory for contextual details: Effects of emotion and neuroimage.2012.03.068
aging. Psychology and Aging, 20(2), 241–250. doi:10.1037/ Mitchell, K. J., & Johnson, M. K. (2009). Source monitoring 15
0882-7974.20.2.241 years later: What have we learned from fMRI about the
Kensinger, E. A., & Schacter, D. L. (2006). Amygdala activity is neural mechanisms of source memory? Psychological
associated with successful encoding of item, but not Bulletin, 135(4), 638–677. doi:10.1037/a0015849
source, information for positive and negative stimuli. The Mitchell, K. J., & Zaragoza, M. S. (2001). Contextual overlap and
Journal of Neuroscience, 26(9), 2564–2570. doi:10.1037/ eyewitness suggestibility. Memory & Cognition, 29, 616–626.
0882-7974.20.2.241 doi:10.3758/BF03200462
Kerns, J. G., Cohen, J. D., MacDonald, A. W., Cho, R. Y., Stenger, Morgan, C. A., III, Southwick, S., Steffian, G., Hazlett, G. A., &
V. A., & Carter, C. S. (2004). Anterior cingulate conflict Loftus, E. F. (2013). Misinformation can influence memory
monitoring and adjustments in control. Science, 303(5660), for recently experienced, highly stressful events.
1023–1026. doi:10.1126/science.1089910 International Journal of Law and Psychiatry, 36, 11–17.
Kim, H., & Cabeza, R. (2007). Trusting our memories: doi:10.1016/j.ijlp.2012.11.002
Dissociating the neural correlates of confidence in veridical Murty, V. P., Ritchey, M., Adcock, R. A., & LaBar, K. S. (2010).
versus illusory memories. The Journal of Neuroscience: The fMRI studies of successful emotional memory encoding: A
Official Journal of the Society for Neuroscience, 27(45), quantitative meta-analysis. Neuropsychologia, 48(12), 3459–
12190–12197. doi:10.1523/JNEUROSCI.3408-07.2007 3469. doi:10.1016/j.neuropsychologia.2010.07.030
Leshikar, E. D., Cassidy, B. S., & Gutchess, A. H. (in press). Nourkova, V., Bernstein, D., & Loftus, E. F. (2004). Altering
Similarity to the self influences cortical recruitment during traumatic memory. Cognition & Emotion, 18(4), 575–585.
impression formation. Cognitive, Affective, & Behavioral doi:10.1080/02699930341000455
Neuroscience. doi:10.3758/s13415-015-0390-3 Odegard, T. N., & Lampinen, J. M. (2004). Memory conjunction
Leshikar, E. D., Dulas, M. R., & Duarte, A. (2015). Self-referen- errors for autobiographical events: More than just familiar-
cing enhances recollection in both young and older adults. ity. Memory, 12, 288–300. doi:10.1080/09658210244000621
Aging, Neuropsychology, and Cognition, 22(4), 388–412. Okado, Y., & Stark, C. (2003). Neural processing associated
doi:10.1080/13825585.2014.957150 with true and false memory retrieval. Cognitive, Affective,
Leshikar, E. D., Park, J. M., & Gutchess, A. H. (2015). Similarity to & Behavioral Neuroscience, 3(4), 323–334. doi:10.3758/
the self affects memory for impressions of others in CABN.3.4.323
younger and older adults. The Journals of Gerontology Okado, Y., & Stark, C. E. (2005). Neural activity during encoding
Series B: Psychological Sciences and Social Sciences, 70(5), predicts false memories created by misinformation.
737–742. doi:10.1093/geronb/gbt132 Learning & Memory, 12(1), 3–11. doi:10.1101/lm.87605
 SOCIAL NEUROSCIENCE 93

Paller, K. A., & Gonsalves, B. (2000). Neural events that underlie attention. Journal of Experimental Social Psychology, 44(5),
remembering something that never happened. Nature 1322–1327. doi:10.1016/j.jesp.2008.03.006
Neuroscience, 3, 1316–1321. doi:10.1038/81851 Vedhara, K., Hyde, J., Gilchrist, I. D., Tytherleigh, M., &
Pinto, I. R., Marques, J. M., Levine, J. M., & Abrams, D. (2010). Plummer, S. (2000). Acute stress, memory, attention and
Membership status and subjective group dynamics: Who cortisol. Psychoneuroendocrinology, 25(6), 535–549.
triggers the black sheep effect? Journal of Personality and doi:10.1016/S0306-4530(00)00008-1
Social Psychology, 99(1), 107–119. doi:10.1037/a0018187 Waring, J. D., & Kensinger, E. A. (2011). How emotion leads to
Schacter, D. L., Buckner, R. L., Koutstaal, W., Dale, A. M., & Rosen, selective memory: Neuroimaging evidence. Neuropsychologia,
B. R. (1997). Late onset of anterior prefrontal activity during 49, 1831–1842. doi:10.1016/j.neuropsychologia.2011.03.007
true and false recognition: An event-related fMRI study. Wilder, D. A. (1990). Some determinants of the persuasive
Neuroimage, 6(4), 259–269. doi:10.1006/nimg.1997.0305 power of in-groups and out-groups: Organization of infor-
Schacter, D. L., Reiman, E., Curran, T., Yun, L. S., Bandy, D., mation and attribution of independence. Journal of
McDermott, K. B., & Roediger, H. L. (1996). Neuroanatomical Personality and Social Psychology, 59(6), 1202–1213.
correlates of veridical and illusory recognition memory: doi:10.1037/0022-3514.59.6.1202
Evidence from positron emission tomography. Neuron, 17 Woo, C.-W., Krishnan, A., & Wager, T. D. (2014). Cluster-extent
(2), 267–274. doi:10.1016/S0896-6273(00)80158-0 based thresholding in fMRI analyses: Pitfalls and recom-
Schacter, D. L., & Slotnick, S. D. (2004). The cognitive neu- mendations. NeuroImage, 91, 412–419. doi:10.1016/j.
roscience of memory distortion. Neuron, 44, 149–160. neuroimage.2013.12.058
doi:10.1016/j.neuron.2004.08.017 Zaragoza, M. S., Belli, R. S., & Payment, K. E. (2006).
Stark, C. E., Okado, Y., & Loftus, E. F. (2010). Imaging the Misinformation effects and the suggestibility of eyewitness
reconstruction of true and false memories using sensory memory. In M. Garry & H. Hayne (Eds.), Do justice and let the
reactivation and the misinformation paradigms. Learning & sky fall: Elizabeth F. Loftus and her contributions to science,
Memory, 17, 485–488. doi:10.1101/lm.1845710 law, and academic freedom (pp. 35–63). Hillsdale, NJ:
Symons, C. S., & Johnson, B. T. (1997). The self-reference effect Lawrence Erlbaum Associates.
in memory: A meta-analysis. Psychological Bulletin, 121, Zaragoza, M. S., & Lane, S. M. (1994). Source misattributions
371–394. doi:10.1037/0033-2909.121.3.371 and the suggestibility of eyewitness memory. Journal of
Trawalter, S., Todd, A. R., Baird, A. A., & Richeson, J. A. (2008). Experimental Psychology: Learning, Memory, and Cognition,
Attending to threat: Race-based patterns of selective 20(4), 934. doi:10.1037/0278-7393.20.4.934
Copyright of Social Neuroscience is the property of Taylor & Francis Ltd and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.