Metacognitive and Metamemory Beliefs in the Development and Maintenance of Posttraumatic Stress Disorder

Abstract

Can metacognition increase trauma sufferers’ risk for developing and maintaining posttraumatic stress disorder (PTSD)? We assessed the role of a range of cognitive and metacognitive belief domains—including metamemory—on PTSD symptoms. Adult participants reported their existing meta/cognitions and lifetime exposure to trauma, then 12 weeks later, they reported meta/cognitions and PTSD symptoms in relation to new trauma exposure since the initial assessment. Participants with more PTSD symptoms held more problematic metacognitions than participants with fewer distress symptoms. Moreover, people who endorsed maladaptive metacognitions before trauma exposure were more likely to experience symptoms of PTSD after exposure. Metacognition predicted the maintenance of elevated PTSD symptoms over the 12-week delay. Our findings support the metacognitive model of PTSD and highlight the importance of metamemory, an understudied factor in PTSD research.

Full text

Clinical Psychological Science
1 –10
© The Author(s) 2016
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/2167702616649348
cpx.sagepub.com
Brief Empirical Report
Exposure to a sudden or sustained stressful experience
can lead to psychological problems such as posttraumatic
stress disorder (PTSD). Hallmark symptoms of PTSD
include repeated and unwanted reexperiencing of the
event, negative alterations in arousal, reactivity, cognition
and mood, and avoidance of trauma reminders (Ameri-
can Psychiatric Association [APA], 2013). Yet, attesting to
human resilience, not all trauma-exposed people develop
PTSD (Lee, 2006). Determining why some trauma-
exposed people develop serious psychopathology when
others do not is of critical clinical significance. Recently,
metacognition—beliefs about thinking that guide our
thinking and coping—has received attention for its role
in PTSD (Wells, 2000). PTSD sufferers who endorse mal-
adaptive metacognitive beliefs posttrauma tend to exhibit
more PTSD symptoms (e.g., Roussis & Wells, 2006). How-
ever, research to date has not examined the role meta-
cognition might play in trauma reactions over time. Here,
we examined whether dysfunctional metacognition
pretrauma predicted PTSD symptomatology posttrauma,
and whether metacognitive beliefs predicted the mainte-
nance of elevated PTSD symptom levels over time.
Wells’s (2000; Wells & Sembi, 2004) metacognitive
model focuses on how people’s metacognitive beliefs
can lead to PTSD. It stipulates that intrusions, startle
responses, and increased arousal are normal responses to
trauma, forming part of a self-righting, reflexive adapta-
tion process (RAP) that initiates automatically and deter-
mines adjustment and recovery. The RAP’s goal is to
simulate plans for future threats. Thus, symptoms should
subside once a satisfactory plan is established. However,
metacognitive beliefs that encourage dysfunctional think-
ing styles or maintain focus on danger or the person’s
649348CPXXXX10.1177/2167702616649348Takarangi et al.Metacognition, Metamemory, and PTSD
research-article2016
Corresponding Author:
Melanie K. T. Takarangi, School of Psychology, Flinders University,
GPO Box 2100, Adelaide, South Australia, 5001
E-mail: melanie.takarangi@flinders.edu.au
Metacognitive and Metamemory Beliefs
in the Development and Maintenance of
Posttraumatic Stress Disorder
Melanie K. T. Takarangi1, Rashelle A. Smith1, Deryn Strange2,
and Heather D. Flowe3
1School of Psychology, Flinders University; 2Department of Psychology, John Jay College of Criminal Justice, City
University of New York; and 3School of Sport, Exercise, and Health Sciences, Loughborough University
Abstract
Can metacognition increase trauma sufferers’ risk for developing and maintaining posttraumatic stress disorder
(PTSD)? We assessed the role of a range of cognitive and metacognitive belief domains—including metamemory—on
PTSD symptoms. Adult participants reported their existing meta/cognitions and lifetime exposure to trauma, then 12
weeks later, they reported meta/cognitions and PTSD symptoms in relation to new trauma exposure since the initial
assessment. Participants with more PTSD symptoms held more problematic metacognitions than participants with
fewer distress symptoms. Moreover, people who endorsed maladaptive metacognitions before trauma exposure were
more likely to experience symptoms of PTSD after exposure. Metacognition predicted the maintenance of elevated
PTSD symptoms over the 12-week delay. Our findings support the metacognitive model of PTSD and highlight the
importance of metamemory, an understudied factor in PTSD research.
Keywords
well-being, trauma, posttraumatic stress disorder, cognitive processes, cognition
Received 6/14/14; Revision accepted 4/14/16
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

2 Takarangi et al.
unsatisfactory reactions to trauma—worry/rumination,
thought suppression, threat monitoring—can obstruct the
RAP and thus interfere with spontaneous recovery from
trauma.
Researchers have investigated a range of maladaptive
cognitive and metacognitive beliefs trauma-exposed peo-
ple hold. We know people who negatively appraise their
traumatic experience are at increased risk of pathology
(Foa, Ehlers, Clark, Tolin, & Orsillo, 1999) and people
who interpret their intrusive memories negatively are less
able to overcome their posttraumatic symptoms (e.g.,
Ehlers, Mayou, & Bryant, 1998). Recent evidence suggests
that training people to adopt a positive appraisal style
regarding their ability to appropriately respond to trauma
leads to fewer analogue symptoms (Woud, Holmes,
Postma, Dalgleish, & Mackintosh, 2012). Moreover, Kleim
etal. (2013) found that changes in dysfunctional trauma-
related appraisals led to decreased PTSD symptoms
among PTSD patients who received trauma-focused cog-
nitive behavior therapy.
Indeed, people who hold maladaptive beliefs pre-
trauma may be predisposed to develop PTSD. Bryant and
Guthrie (2005) found that trainee firefighters with a pre-
existing tendency for negative self-appraisal—but not
cognition concerning self-blame or the world as unsafe—
were more symptomatic 6 months later. These results
suggest that studying people’s cognition—and perhaps
their metacognition—pretrauma may help identify peo-
ple most at risk for PTSD symptomatology.
Bennett and Wells (2010) found student nurses who
endorsed negative metamemory beliefs (“having gaps in
memory of the event means I am not normal”) were
more likely to exhibit PTSD symptoms after a distressing
event during their training. Moreover, such beliefs pre-
dicted PTSD better than objective indicators of memory
problems in participants’ recall narratives. These data
suggest that metacognitive beliefs about memorial prob-
lems may be an important area for further research.
In summary, extant research demonstrates that dys-
functional meta/cognition may render people more vul-
nerable to increased PTSD symptomatology. Does
metacognition also play a role in maintaining posttrau-
matic stress? Some studies show that posttrauma cogni-
tion independently predicts PTSD when measured
between several months and 1 year posttrauma (e.g.,
Ehlers et al., 1998; Halligan, Michael, Clark, & Ehlers,
2003) and, in children, mediates the relationship between
initial and longer term PTSD symptoms (Meiser-Stedman,
Dalgleish, Glucksman, Yule, & Smith, 2009). However, to
our knowledge, no study has investigated whether meta-
cognition contributes to the maintenance of persistent
PTSD symptoms in adults over time.
The Present Study
We investigated the role of meta/cognitive factors in pre-
dicting and maintaining PTSD among a nonclinical adult
population. Even short research time frames can capture
a high prevalence of trauma in nonclinical populations
(e.g., 8 weeks; Frazier etal., 2009). However, to increase
the likelihood that we would capture trauma, we assessed
participants over 12 weeks. We measured participants’
trauma-related cognition, metacognition, and PTSD
symptoms pre– and post–any recent trauma exposure.
Our aims were threefold. First, we examined the cross-
sectional relationship between a range of cognitive and
metacognitive belief domains—including positive and
negative metamemory beliefs—and PTSD reactions to
traumatic events. Second, we examined whether preex-
isting metacognitive beliefs (Time 1 [T1]) increased PTSD
symptomatology after trauma (Time 2 [T2]). Third, we
investigated whether metacognition predicted the main-
tenance of elevated PTSD symptom levels over time.
Method
Participants
Participants (N = 664) were residents of Australia (n =
74), Canada (n = 17), New Zealand (n = 4), the United
Kingdom (n = 51), and the United States (n = 518), at
least 18 years old, and fluent in English. We recruited
participants from Amazon Mechanical Turk (n = 372;
they received $0.75 at T1 and $0.75 at T2, in keeping with
rates of compensation by time for other psychology stud-
ies), the Flinders University Psychology Research Partici-
pation Pool (n = 33, course credit), and standard research
recruitment and social networking sites (n = 259, volun-
tary participation). Of the 683 who completed the T1
survey, 315 also participated at T2. We embedded manip-
ulation checks to ensure participants paid attention and
excluded participants who failed more than one (T1: n =
13, T2: n = 10). From T1, we also excluded 5 participants
who did not meet the eligibility criteria and 1 who com-
pleted the survey twice. From T2, we excluded 1 partici-
pant who had failed attention checks at T1, 1 who could
not be matched to his or her T1 data, and 3 who com-
pleted T2 18 weeks or more after T1. The final sample
comprised 664 T1 participants and 300 T2 participants.
Table 1 shows their demographics and descriptive infor-
mation for the main measures. Notably, at T1, participants
who did not complete T2 were older (p < .01), reported
more PTSD symptoms (p < .01), and had higher problem-
atic cognition and metacognition on all but positive
metamemory beliefs (ps < .05), compared with completers.
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

3
Table 1. Demographic, Trauma-Related Characteristics, and Main Measures for the Total Sample Including Cross-Sectional Pearson’s
Correlation Coefficients and Confidence Intervals With PTSD Symptoms
Time 1 (n = 664) Time 2 (n = 300) Time 1 (n = 664) Time 2 (n = 300)
Characteristic M (SD), R M (SD), R r [CI] With PCL r [CI] With PCL
Demographics
Age in years 32.38 (12.34), 18–75 35.32 (12.96), 18–75
Female sex, % (n) 68.5 (455) 67.7 (203)
Ethnicity
Caucasian, % (n) 83.4 (554) 85.0 (255)
Minority, % (n)a16.6 (110) 14.7 (42)
Level of education achieved
School, % (n) 31 (206) 28 (84)
College/university, % (n) 65.5 (435) 68.7 (206)
TAFE/trade, % (n) 3.5 (23) 3.3 (10)
Employment
Employed, % (n) 63.2 (420) 70.7 (212)
Unemployed, % (n) 16.7 (111) 15.6 (47)
Student, % (n) 20 (133) 13.7 (41)
Trauma characteristics and symptoms
Age at trauma 20.80 (11.56), 0–70 35.32 (12.96), 18–75 −.19 [−.27, −.11] *** −.18 [−.29, .07] **
Time since traumab12.12 (11.44), 0−57 26.04 (22.02), 0−105 −.08 [−.16, −.01] .05 [−.15, 26]
Number of traumas 4.06 (2.72), 0–13 4.36 (2.99), 0–15 .38 [.31, .44] *** .41 [.32, .51] ***
Distress at trauma 4.59 (.79), 1–5 3.93 (1.38), 1–5 .21 [.14, .28] *** . 43 [.26, .58] ***
HADS total 16.67 (7.06), 0–42 12.80 (8.29), 0–42 .68 [.64, .72] *** .74 [.70, .77] ***
HADS anxiety 10.40 (4.25), 0–21 7.61 (4.62), 0–21 .64 [.59, .70] *** .70 [.66, .74] ***
HADS depression 6.27 (3.63), 0–21 5.18 (4.52), 0–21 .56 [.49, .62] *** .64 [.59, .68] ***
PCL total 37.09 (17.38), 17–85 32.52 (14.97), 17–84 — —
Metacognition measures
BAMQ positive 14.38 (6.27), 8 – 32 15.34 (6.75), 9–36 .44 [.38, .50] *** .56 [.48, .63] ***
BAMQ negative 10.01 (3.98), 7–28 10.72 (3.98), 8–32 .58 [.52, .62] *** .60 [.52, .67] ***
MCQ-30 positive worry 10.84 (4.35), 6–24 10.60 (4.10), 6–24 .24 [.17, .31] *** .30 [.19, .40] ***
MCQ-30 uncontrollability 13.2 (5.17), 6–24 12.29 (4.87), 6–24 .54 [.48, .59] *** .53 [.44, .61] ***
MCQ-30 thought control 11.84 (4.03), 6–24 11.04 (4.14), 6–24 .43 [.37, .49] *** .39 [.29, .48] ***
PTCI self 2.65 (1.42), 1–7 2.46 (1.37), 1–7 .71 [.67, .74] *** .76 [.71, .80] ***
PTCI world 4.23 (1.65), 1–7 3.92 (1.69), 1–7 .57 [.51, .61] *** .56 [.48, .63] ***
PTCI self-blame 2.53 (1.58), 1–7 2.32 (1.56), 1–7 .46 [.40, .52] *** .43 [.33, .52] ***
RIQ intrusions 14.81 (9.36), 6–42 13.86 (8.79), 6–42 .70 [.64, .74] *** .76 [.69, .81] ***
Note: In the first two columns, values are M (SD), R, unless otherwise noted. BAMQ = Beliefs About Memory Questionnaire; CI = confidence interval; HADS =
Hospital Anxiety and Depression Scale; MCQ-30 = Metacognitions Questionnaire–30; PCL = PTSD Checklist–Specific Version; PTCI = Posttraumatic Cognitions
Inventory; PTSD = posttraumatic stress disorder; RIQ = Response to Intrusions Questionnaire; TAFE = Technical and Further Education.
aMinority, e.g., African American, Hispanic, Middle Eastern, Asian American, multiracial. There were no significant differences in PTSD symptoms between
White and non-White participants at either time point, and we therefore excluded ethnicity from further analysis. bTime since trauma = years at T1, days at T2.
**p < .01. ***p < .001.
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

4 Takarangi et al.
This research was approved by the Social and Behav-
ioural Research Ethics Committee at Flinders University
and conducted in accordance with the provisions of the
World Medical Association Declaration of Helsinki.
Measures
We administered all measures at T1 and T2, 12 weeks
apart.
Personal information. Participants completed demo-
graphic questions (age, ethnicity, country of residence,
highest education level achieved) and provided their
email address.
Depression and anxiety. The 14-item Hospital Anxi-
ety and Depression Scale (Zigmund & Snaith, 1983)
assessed anxiety and depression symptoms. Participants
rated each item from 0 to 3 according to how they felt
during the previous week (anchors vary by item). The
Depression subscale focuses mainly on the reduced plea-
sure response aspect of depression (“I feel cheerful”; 0 =
most of the time, 3 = not at all), whereas the Anxiety sub-
scale focuses on generalized anxiety and panic (“worry-
ing thoughts go through my mind”; 0 = only occasionally,
3 = a great deal of the time). For our sample, internal
consistency was as follows: Anxiety T1 = .81, T2 = .88;
Depression T1 = .73, T2 = .87.
Traumatic events. The Trauma History Screen (Carlson
et al., 2011) assessed lifetime exposure to potentially
traumatic events at T1; at T2, participants indicated
whether they had experienced any of the events since
T1. Participants also specified their age (or days since the
event for T2) and how emotionally distressed they were
at the time of the event (not at all, a little, somewhat,
much, very much). Next, participants nominated their
worst event and described that event briefly. The psycho-
metric properties of this scale are comparable or better
than longer measures of trauma exposure (Carlson etal.,
2011).
Posttraumatic stress symptoms. The PTSD Check-
list–Specific Version (PCL; Weathers, Litz, Herman, Huska,
& Keane, 1993) assessed posttraumatic stress. Participants
responded to 17 items regarding their self-nominated
worst event at T1 and T2 and their symptoms within the
past 2 weeks, using a 5-point scale (1 = not at all, 5 =
extremely). Internal consistency in the current sample
was .95 at T1 and T2. To investigate symptom mainte-
nance over the T1 to T2 delay, we classified participants’
PTSD symptoms as “persistent” if they displayed elevated
PCL scores (>44) at T1 and T2 and “recovered” if they
displayed elevated scores at T1 but not at T2 (i.e., we
followed previous research: Blanchard, Jones-Alexander,
Buckley, & Forneris, 1996; Bonanno, 2005; Robinaugh
etal., 2011).
Trauma-related cognitions. The Posttraumatic Cogni-
tions Inventory (PTCI; Foa etal., 1999) measured nega-
tive cognitions about the self (“I can’t rely on myself”)
and the world (“you can never know what or who may
harm you”) and self-blame (“the event happened because
of the way I acted”). Participants rated their agreement
with each statement (1 = totally disagree, 7 = totally
agree). Here, the PTCI demonstrated adequate reliability
for each subscale (Self: .96 and .97; World: .92 and .91;
Self-Blame: .88 and .94).
Metacognitive beliefs. We selected three subscales of
the Metacognitions Questionnaire–30 (MCQ-30; Wells &
Cartwright-Hatton, 2004) due to their established associa-
tion with PTSD (Roussis & Wells, 2006): positive beliefs
about worry (“worrying helps me cope”), beliefs about
the uncontrollability and danger of thoughts (“my worry-
ing is dangerous for me”), and beliefs about the need to
control thoughts (“it is bad to think certain thoughts”).
Participants rated their agreement with each statement on
a 4-point scale (1 = do not agree, 4 = agree very much).
The MCQ-30 is well validated in adults (Wells &
Cartwright-Hatton, 2004). Cronbach’s alpha ranged from
.77 to .92. The 10-item Response to Intrusions Question-
naire (RIQ; Clohessy & Ehlers, 1999) measured negative
inferences about the meaning of intrusive memories.1
Participants who had experienced intrusions rated what
they thought their intrusions meant (“my life is ruined”)
on a 7-point scale (1 = totally disagree, 7 = totally agree).
The scale has previously shown good reliability and pre-
dictive validity (e.g., Clohessy & Ehlers, 1999; Halligan
etal., 2003). Here, internal consistency was .92 at T1 and
.91 atT2.
Metamemory. The Beliefs About Memory Question-
naire (BAMQ; Bennett & Wells, 2010) assessed metacog-
nition about traumatic memory. The 15 items measured
positive beliefs about needing a complete trauma mem-
ory (“I must try to remember all of the details of the
event so that I can understand why it happened”) and
negative beliefs about the consequences of not having a
complete memory (“gaps in my memory for the event
are preventing me from getting over it”). Participants
rated their agreement with each belief on a 4-point scale
(1 = do not agree, 4 = agree very much). Bennett and
Wells (2010) found preliminary support for the conver-
gent validity of the BAMQ subscales with subscales of
the MCQ-30. Here, the internal consistency was as fol-
lows: Positive Beliefs: T1 = .91, T2 = .92; Negative Beliefs:
T1 and T2 = .85.
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

Metacognition, Metamemory, and PTSD 5
Because responses on the PTCI, BAMQ, and RIQ are
anchored to a specific traumatic event, we asked partici-
pants to respond in reference to their worst event or, if
they had not described one, a negative event they had
experienced.
Procedure
At T1, participants completed the questionnaire battery
online. Approximately 12 weeks later, we emailed partici-
pants who consented to be recontacted with the link to
the T2 survey (completion time = 77–91 days after T1).
Participants received full debriefing information at the
study conclusion.
Results
All analyses were two-tailed and alpha was set at .05.
Several subscale scores at T1 and T2 were positively
skewed. Although square root and log transformations
reduced the skew, the overall pattern of results was the
same; thus, we retained the original untransformed data
for analysis. We replaced missing items on questionnaires
using mean substitution (by subscale).
Consistent with prior studies, most (n = 633, 95.3%)
participants reported experiencing at least one lifetime
trauma (Breslau et al., 1998; Frazier et al., 2009).2 The
most frequent self-nominated worst events were the
unexpected death of a loved one (n = 173), “other”
trauma (n = 105, e.g., childhood emotional abuse, still-
births, kidnappings), child sexual assault (n = 67), sud-
den abandonment (n = 48), and transport accident (n =
46). At T2, 35% (n = 107) reported experiencing at least
one “new” trauma; all but one also reported a lifetime
trauma at T1. Of these recent events, “other” events were
most commonly nominated as the worst (n = 48, e.g.,
elementary school shootings, Boston bombings, Hurri-
cane Sandy). Of participants displaying elevated (>44)
PTSD symptoms at T1 (n = 214, of whom 78 also com-
pleted T2), 56.4% (n = 44) showed persistently elevated
PTSD symptoms at T2, and 43.6% (n = 34) were classified
as recovered.
Table 1 shows the cross-sectional relationships among
the demographic, cognitive, metacognitive, and symp-
tom variables at T1 and T2. Females reported signifi-
cantly more PTSD symptoms than males at T1, t(662) =
−2.84, p < .01, d = .24. Age at trauma was negatively
related to PTSD. Number of prior traumas, emotional
distress at the time of the trauma, anxiety, depression,
and all types of cognition and metacognition were posi-
tively associated with PTSD symptoms at both time
points. Participants’ negative cognitions about the self
were highly correlated with other variables; in particular
with PTSD (r = .70) and RIQ (r = .83). Thus, we excluded
this variable from our regression analyses due to
multicollinearity.3
We next conducted a (forced entry) hierarchical
regression to examine whether certain demographic,
cognitive, and metacognitive variables predicted the
degree of concurrent PTSD symptomatology at T1.4 In
Step 1, we entered five control variables—age at trauma,
distress at the time of recent worst trauma exposure, gen-
der, number of prior traumas, and depression; these vari-
ables accounted for a significant proportion of the
variance in symptoms (43%). All were independent pre-
dictors except gender and distress; depression, which is
often comorbid with PTSD (APA, 2013), was the strongest
predictor. In Step 2, preexisting cognitive beliefs
accounted for a significant additional proportion of
symptom variance (10%); depression, age at trauma, and
number of prior traumas remained significant predictors,
and world beliefs and self-blame were also independent
predictors. In Step 3, metacognitive beliefs also accounted
for a significant proportion of symptom variance over
and above the control and cognitive variables (11%).
Depression, prior traumas, and world beliefs from Step 2
remained significant predictors, along with beliefs con-
cerning the uncontrollability and danger of thoughts,
positive and negative metamemory beliefs, and negative
inferences about intrusive memories. The overall model
explained 64% total variance in PTSD, F(13, 348) = 47.96,
p < .001.
We next examined whether preexisting metacognitive
beliefs predicted degree of PTSD symptomatology fol-
lowing recent trauma exposure (i.e., exposure between
T1 and T2); we conducted a (forced entry) hierarchical
regression. Here, we included only participants who had
experienced a novel trauma (n = 107) between T1 and
T2, with T2 PCL as the outcome variable. Of these, 25
participants had no score for the RIQ, because it was
administered only to participants who reported intru-
sions. To maximize the available sample size, we used
the Expectation-Maximization technique in SPSS to
replace missing values for participants who did not report
age at the time of the trauma (n = 3), distress relating to
the trauma (n = 7), or either of these variables (n = 2).
We entered six control variables—age, distress at the time
of recent trauma exposure, depression, gender, number
of prior traumas, PTSD symptoms at T1—simultaneously
in Step 1. We added the cognitive variables (T1) in Step
2 and the metacognitive variables (T1) in Step 3.
Table 2 displays the regression statistics for each step
of the model. As shown in Step 1, the control variables
accounted for a significant proportion of variance in T2
PTSD symptoms (58%). In Step 2, preexisting cognitive
beliefs did not explain significant additional variance
(1%). However, as predicted, in Step 3 preexisting meta-
cognitive beliefs did account for a significant proportion
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

6 Takarangi et al.
of additional variance (10%). In particular, preexisting
beliefs concerning the uncontrollability/danger of
thoughts and negative inferences about intrusive memo-
ries independently predicted PTSD symptoms after
recent trauma. These variables are important contribu-
tors to the final model, uniquely accounting for 3.84%
and 3.24% of the variance, respectively. Other critical
variables are depression (4.08%) and prior traumas
(2.46%). However, contrary to our hypothesis, people
who, at T1, were less likely to believe the world is dan-
gerous and that they needed to control thoughts had
more symptoms of PTSD at T2 (2.82%, 1.99%). The over-
all model explained 69% total variance in PTSD, F(14,
67) = 10.77, p < .001.
Table 2. Summary of Hierarchical Regression of Control Variables, Cognitive Beliefs,
and Metacognitive Beliefs at Time 1 in Predicting Posttraumatic Stress Symptoms at
Time 2
Predictor B SE β
Squared
semipartial
correlation VIF
Step 1
Constant −1.99 7.82
Age at time of trauma −0.04 0.10 −.03 .001 1.04
Number of traumas 0.38 0.46 .06 .004 1.19
Distress at trauma 3.69 1.59 .19 * .030 1.25
Gender −0.82 3.13 −.02 <.001 1.29
Time 1 PTSD 0.43 0.10 .45 *** .111 1.82
Time 1 depression 1.44 0.47 .30 ** .052 1.75
R2 = .58, F(6, 75) = 17.53, p < .001
Step 2
Constant 3.52 8.70
Age at time of trauma −0.04 0.10 −.03 .001 1.05
Number of traumas 0.54 0.47 .10 .007 1.25
Distress at trauma 3.07 1.65 .16 .019 1.35
Gender −0.23 3.14 −.01 <.001 1.31
Time 1 PTSD 0.49 0.11 .52 *** .112 2.39
Time 1 depression 1.68 0.49 .35 ** .064 1.94
PTCI world −1.89 1.23 −.16 .013 1.97
PTCI self-blame −0.09 0.97 −.01 <.001 1.68
R2Change = .01, FChange(2, 73) = 1.26, p = .289
Step 3
Constant 3.51 8.91
Age at time of trauma −0.00 0.10 −.00 <.001 1.27
Number of traumas 1.03 0.44 .18 * .025 1.34
Distress at trauma 2.02 1.62 .11 .007 1.56
Gender 0.40 3.15 .01 <.001 1.59
Time 1 PTSD 0.25 0.12 .26 * .020 3.37
Time 1 depression 1.46 0.49 .31 ** .041 2.30
PTCI world −2.89 1.17 −.25 * .028 2.14
PTCI self-blame −0.40 0.94 −.04 .001 1.89
RIQ intrusions 0.44 0.17 .28 * .032 2.36
BAMQ positive metamemory 0.20 0.22 .08 .004 1.68
BAMQ negative metamemory 0.14 0.38 .04 .001 2.39
MCQ-30 positive worry 0.15 0.30 .04 .001 1.36
MCQ-30 uncontrollability/danger 1.04 0.36 .31 ** .038 2.52
MCQ-30 control thoughts −0.84 0.41 −.22 * .020 2.48
R2Change = .10, FChange(6, 67) = 3.43, p = .005
Note: BAMQ = Beliefs About Memory Questionnaire; MCQ-30 = Metacognitions Questionnaire–30;
PTCI = Posttraumatic Cognitions Inventory; PTSD = posttraumatic stress disorder; RIQ = Response
to Intrusions Questionnaire; VIF = variance inflation factor.
*p < .05. **p < .01. ***p < .001.
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

Metacognition, Metamemory, and PTSD 7
We also predicted that maladaptive meta/cognitions
would drive the persistence of PTSD and hence differenti-
ate people who experienced elevated PTSD symptoms
between T1 and T2 and people whose symptoms abated.
Using only those participants whose symptoms were
classified as “persistent” or “recovered” at T2 (n = 78, less
12 participants with no RIQ score), we entered the cogni-
tive and metacognitive variables into a (forced entry)
hierarchical multivariate logistic regression analysis.5 This
model accurately predicted 83.3% of cases and contained
only negative metamemory beliefs (OR = 1.35 [1.01, 1.81],
p < .05) and negative inferences about the meaning of
intrusions (OR = 1.17 [1.03, 1.33], p = .01) as significant
predictors of persistent PTSD symptoms (χ2 = 38.61,
p<.01).
Discussion
We examined the relationship between specific types of
metacognition and the development and maintenance of
posttraumatic stress symptoms. Considering the correla-
tional data, at T1 and T2, the more our participants held
unhelpful beliefs, the more PTSD symptomatology they
showed. It is interesting that cognition about the self was
the most important correlate of PTSD symptomatology,
corroborating previous research (vs. world/self-blame;
Bryant & Guthrie, 2005; Foa & Rauch, 2004; Moser,
Hajcak, Simons, & Foa, 2007). This result is not surprising,
given substantial conceptual overlap between negative
self-related cognition and symptoms, as well as with our
other key variables. Of the metacognitive beliefs, at T1
negative metamemory beliefs were the most important,
whereas at T2 it was beliefs about intrusions. However,
these particular results do not speak to whether pretrauma
metacognition predicts later PTSD symptomatology.
Thus, we next examined whether cognitive and meta-
cognitive factors predicted PTSD symptomatology after
trauma exposure. We found that preexisting cognitive
and metacognitive beliefs (T1) predicted PTSD symptom
levels after exposure to a novel trauma (T2). Specifically,
people reported more PTSD symptoms when exposed to
trauma if they had showed concern that their thoughts
were dangerous and made negative inferences about the
meaning of their intrusive symptoms. In addition, people
who, pretrauma, believed the world to be more safe and
predictable, and did not believe that they should control
their thoughts, were more likely to report more symp-
toms posttrauma. Perhaps for people with overly optimis-
tic views of the safety and fairness of the world, trauma
shatters their basic beliefs and leads them to shift those
beliefs to the other extreme (Foa & Riggs, 1993; Janoff-
Bulman, 1992). Our data, like those of Bryant and
Guthrie (2005), suggest beliefs that the world is danger-
ous and that it is bad to think certain thoughts
are activated following—rather than prior to—trauma
exposure (cf. Foa & Riggs, 1993). Finally, we found the
extent to which people held metamemory beliefs that
fragmentary trauma memory was pathological and made
negative inferences about the meaning of intrusions
played a significant role in maintaining elevated PTSD
symptoms over 12 weeks. These results support the
hypothesis that metacognition plays an important role in
PTSD maintenance among adults.
Taken together, our results support the metacognitive
model of PTSD (Wells, 2000): Problematic metacogni-
tions likely blocked participants’ adaptation process,
increasing their focus on threat such that anxiety and a
sense of ongoing danger persisted, which in turn main-
tained symptoms (e.g., Roussis & Wells, 2006). Our data
also contribute to a growing literature showing the
importance of metamemory beliefs (about intrusive
memories and memory deficits) to PTSD symptoms. In
particular, negative beliefs that a fragmented memory
means something bad, or is abnormal, predicted PTSD
maintenance.
Whether people with PTSD have fragmentary and dis-
organized memories for their traumatic experience has
long been debated (e.g., Ehlers, Ehring, & Kleim, 2012).
Our data suggest that negative metacognition about trau-
matic memory might contribute to trauma-exposed peo-
ple’s struggle to resolve their symptoms. It is interesting
that people’s metacognition becomes more apparent the
more they report recurrently thinking about an event.
The more we think about an event, the more we might
think details are missing from memory. Recovering previ-
ously inaccessible details creates an impression of partial
amnesia (e.g., Berntsen & Rubin, 2014). Yet, nonclinical
research suggests this is not unique to traumatic memory
(Read & Lindsay, 2000). Thus, it is unclear whether disor-
ganization in traumatic memories contributes signifi-
cantly to PTSD symptoms, or if metamemory beliefs and
strategies concerning trauma memory (e.g., rumination,
“gap filling”) play a greater role in maintaining symp-
toms. We do know, however, that rumination mediates
the relationship between metamemory beliefs and intru-
sive PTSD symptoms (Bennett & Wells, 2010).
There are several limitations to our study. First,
although we gathered data prospectively, the study lasted
only 12 weeks. In addition, because not all participants
met the one-month duration criterion for PTSD with
symptoms at T2, it is possible that, among this group,
symptomatology represented normal stress reactions to
trauma. It is important that we did not include a clinical
diagnostic interview to assess PTSD or any other psycho-
pathology at T1 that could potentially account for our
results. Also, we used a convenience sample of students
and a broad section of Internet users, who were pre-
dominantly Caucasians from the United States, which
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

8 Takarangi et al.
may limit the generalizability of our findings. Moreover,
because the participants reporting the highest levels of
problematic metacognition and PTSD symptoms at T1
were less likely to complete the entire study, our final T2
sample was not representative of those at the more
extreme end. Also, although we initially recruited a large
sample, due to incomplete data on some measures, the
final sample size we used for our main regression analy-
ses was limited. Finally, in interpreting the separate con-
tribution of metacognitions to the development and
maintenance of posttraumatic stress symptoms, it is
important to acknowledge that the RIQ could potentially
also be indexing cognitions, rather than just pure
metacognitions.
Considering our key findings, and the limitations of
our study, we would suggest two future research areas.
To advance our understanding of how metacognitive
beliefs may lead to PTSD after new trauma exposure, we
suggest conducting a similar longitudinal study with peo-
ple at high risk of trauma exposure, such as emergency
service personnel (e.g., Bryant & Guthrie, 2005). Research
could also examine whether cognitive bias training
(Woud et al., 2012) directed specifically at changing
problematic metacognitive beliefs halts the development
of symptoms after new trauma or reduces symptoms
relating to previous trauma. This method would help
determine whether metacognitive beliefs are causal risk
factors for PTSD (Kraemer etal., 1997). In both lines of
research, we suspect it will be important to employ a
longer delay when assessing the role of metacognition in
maintaining PTSD symptoms over time and to attempt to
capture metacognition and symptoms immediately after
trauma exposure.
In summary, our results indicate that metacognition—
particularly regarding traumatic memory, a new area of
investigation—plays an active role in adult PTSD. Specifi-
cally, this evidence adds to growing empirical support
that metacognitive beliefs are important to predicting
PTSD symptomatology.
Author Contributions
M. K. T. Takarangi and D. Strange developed the study con-
cept. All authors contributed to the study design. R. A. Smith
performed data collection; R. A. Smith performed initial analy-
ses and interpretation; M. K. T. Takarangi and H. D. Flowe
performed additional analysis. R. A. Smith and M. K. T. Taka-
rangi drafted the manuscript, and D. Strange and H. D. Flowe
provided critical revisions. All authors approved the final ver-
sion of the manuscript for submission.
Acknowledgments
We thank Caitlin Lloyd, Evan Dawson, and Hannah James for
their assistance in preparing study materials, Jacinta Oulton and
Ella Moeck for their assistance with data preparation, and James
Rockey for his advice on ridge regression.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with
respect to their authorship or the publication of this article.
Funding
R.A.S. was supported by an award from the Nursing and Allied
Health Scholarship and Support Scheme (NAHSSS) funded by
the Commonwealth Department of Health and Ageing. The
views expressed in this research do not necessarily represent
those of the NAHSSS, its Administrator, Services for Australian
Rural and Remote Allied Health (SARRAH); and/or the Govern-
ment Department of Health and Ageing. The planning phase of
this research was supported by a Wellcome Trust Biomedical
Vacation Scholarship supervised by M.K.T.T. and H.D.F.
Supplemental Material
Additional supporting information may be found at http://
cpx.sagepub.com/content/by/supplemental-data.
Notes
1. An initial coding error on the survey meant we were unable
to use T1 RIQ data for 92 participants.
2. Table S1 in the Supplemental Material available online shows
the lifetime prevalence of each potentially traumatic event and
participants’ exposure to these events between T1 and T2.
3. We tested our hypotheses using ordinary least squares (OLS)
regression and logistic regression analyses. To address the con-
cern that the predictors in our models could be highly corre-
lated, we verified each analysis using ridge regression (Hoerl &
Kennard, 1970; Millsap & Maydeu-Olivares, 2009). Ridge regres-
sion improves a model’s predictive accuracy in cases where
there are more predictors than observations and/or when the
predictors are highly correlated. The ridge regression results did
not differ from those obtained with OLS and logistic regression;
hence, we report only the latter ones for ease of interpretation.
4. Table S2 in the Supplemental Material available online dis-
plays the full regression table.
5. With the available sample size reduced, here we included
only the theoretical predictors of interest (i.e., meta/cognitive
factors). Table S3 in the Supplemental Material available online
displays the full regression table.
References
American Psychiatric Association. (2013). Diagnostic and sta-
tistical manual of mental disorders. (5th ed.). Washington,
DC: Author.
Bennett, H., & Wells, A. (2010). Metacognition, memory disor-
ganisation and rumination in posttraumatic stress symptoms.
Journal of Anxiety Disorders, 24, 318–325. doi:10.1016/
j.janxdis.2010.01.004
Berntsen, D., & Rubin, D. C. (2014). Involuntary memories
and dissociative amnesia: Assessing key assumptions in
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

Metacognition, Metamemory, and PTSD 9
PTSD research. Clinical Psychological Science, 2, 174–186.
doi:10.1177/2167702613496241.
Blanchard, E. B., Jones-Alexander, J., Buckley, T. C., & Forneris,
C. A. (1996). Psychometric properties of the PTSD Checklist
(PCL). Behaviour Research and Therapy, 34, 669–673.
doi:10.1016/0005-7967(96)00033-2
Bonanno, G. A. (2005). Resilience in the face of potential
trauma. Current Directions in Psychological Science, 14,
135–148. doi:10.1111/j.0963-7214.2005.00347.x
Breslau, N., Kessler, R. C., Chilcoat, H. D., Schultz, L. R., Davis,
G. C., & Andreski, P. (1998). Trauma and posttraumatic
stress disorder in the community: The 1996 Detroit area
survey of trauma. Archives of General Psychiatry, 55, 626–
632. doi:10.1001/archpsyc.55.7.626
Bryant, R. A., & Guthrie, R. M. (2005). Maladaptive appraisals
as a risk factor for posttraumatic stress: A study of trainee
firefighters. Journal of Psychological Science, 16, 749–752.
doi:10.1111/j.1467-9280.2005.01608.x
Carlson, E. B., Smith, S. R., Palmieri, P. A., Dalenberg, C., Ruzek,
J. I., Kimerling, R., . . . Spain, D. A. (2011). Development
and validation of a brief self-report measure of trauma expo-
sure: The Trauma History Screen. Psychological Assessment,
23, 463–477. doi.org/10.1037/a0022294
Clohessy, S., & Ehlers, A. (1999). PTSD symptoms, response to
intrusive memories and coping in ambulance service work-
ers. British Journal of Clinical Psychology, 38, 251–265.
doi:10.1348/014466599162836
Ehlers, A., Ehring, T., & Kleim, B. (2012). Information process-
ing in posttraumatic stress disorder. In J. G. Beck & D. M.
Sloan (Eds.), The Oxford handbook of traumatic stress dis-
orders (pp. 191–218). Oxford, England: Oxford University
Press.
Ehlers, A., Mayou, R. A., & Bryant, B. (1998). Psychological pre-
dictors of posttraumatic stress disorder after motor vehicle
accidents. Journal of Abnormal Psychology, 107, 508–519.
doi:10.1037//0021-843X.107.3.508
Foa, E. B., Ehlers, A., Clark, D. M., Tolin, D. F., & Orsillo, S.M.
(1999). The Posttraumatic Cognitions Inventory (PTCI):
Development and validation. Journal of Psychological
Assessment, 11, 303–314. doi:10.1037/1040-3590.11.3.303
Foa, E. B., & Rauch, S. M. (2004). Cognitive changes during pro-
longed exposure versus prolonged exposure plus cognitive
restructuring in female assault survivors with posttrau-
matic stress disorder. Journal of Consulting and Clinical
Psychology, 72, 879–884. doi:10.1037/0022-006X.72.5.879
Foa, E. B., & Riggs, D. S. (1993). Post-traumatic stress disorder
in rape victims. In J. Oldham, M. B. Riba, & A. Tasman
(Eds.), Annual review of psychiatry (Vol. 12, pp. 273–303).
Washington, DC: American Psychiatric Association.
Frazier, P., Anders, S., Perera, S., Tomich, P., Tennen, H., Park,
C., & Tashiro, T. (2009). Traumatic events among under-
graduate students: Prevalence and associated symptoms.
Journal of Counseling Psychology, 56, 450–460. doi:10.1037/
a0016412
Halligan, S. L., Michael, T., Clark, L. M., & Ehlers, A. (2003).
Posttraumatic stress disorder following assault: The role
of cognitive processing, trauma memory, and appraisals.
Journal of Consulting and Clinical Psychology, 71, 419–
431. doi:10.1037/0022-006X.71.3.419
Hoerl, A. E., & Kennard, R. W. (1970). Ridge regression: Biased
estimation for nonorthogonal problems. Technometrics, 12,
55–67. doi:10.1080/00401706.1970.10488634
Janoff-Bulman, R. (1992). Shattered assumptions: Toward a new
psychology of trauma. New York, NY: Free Press.
Kleim, B., Grey, N., Wild, J., Nussbeck, F. W., Stott, R.,
Hackmann, A., . . . Ehlers, A. (2013). Cognitive change
predicts symptom reduction with cognitive therapy for
posttraumatic stress disorder. Journal of Consulting and
Clinical Psychology, 81, 383–393. doi:10.1037/a0031290
Kraemer, H. C., Kazdin, A. E., Offord, D. R., Kessler, R. C.,
Jensen, P. S., & Kupfer, D. J. (1997). Coming to terms with
the terms of risk. Archives of General Psychiatry, 54, 337–
343. doi:10.1001/archpsyc.1997.01830160065009
Lee, D. (2006). Case conceptualisation in complex PTSD:
Integrating theory with practice. In D. H. Barlow (Ed.),
Clinical handbook of psychological disorders: A step by
step treatment manual (pp. 188–215). London, England:
Guilford.
Meiser-Stedman, R., Dalgleish, T., Glucksman, E., Yule, W., &
Smith, P. (2009). Maladaptive cognitive appraisals mediate
the evolution of posttraumatic stress reactions: A 6-month
follow-up of child and adolescent assault and motor vehi-
cle accident survivors. Journal of Abnormal Psychology,
118, 778–787. doi:10.1037/A0016945
Millsap, R. E., & Maydeu-Olivares, A. (Eds.). (2009). The Sage
handbook of quantitative methods in psychology. London,
England: Sage.
Moser, J. S., Hajcak, G., Simons, R. F., & Foa, E. B. (2007). PTSD
symptoms in trauma-exposed college students: The role of
negative cognitions, trauma type, and anxiety. Journal of
Anxiety Disorders, 21, 1039–1049. doi:10.1016/j.janxdis
.2006.10.009
Read, J. D., & Lindsay, D. S. (2000). “Amnesia” for summer camps
and high school graduation: Memory work increases reports
of prior periods of remembering less. Journal of Traumatic
Stress, 13, 129–147. doi:10.1023/A:1007781100204
Robinaugh, D. J., Marques, L., Traeger, L. N., Marks, E. H., Sung,
S. C., Beck, J. G., . . . Simon, N. M. (2011). Understanding
the relationship of perceived social support to post-trauma
cognitions and posttraumatic stress disorder. Journal of
Anxiety Disorders, 25, 1072–1078. doi:10.1016/j.janxdis
.2011.07.004
Roussis, P., & Wells, A. (2006). Posttraumatic stress symptoms:
Tests of relationships with thought control strategies and
beliefs as predicted by the metacognitive model. Personality
and Individual Differences, 40, 111–112. doi:10.1016/j
.paid.2005.06.019
Weathers, F. W., Litz, B. T., Herman, D. S., Huska, J. A., & Keane,
T. M. (1993, October). The PTSD Checklist: Reliability, valid-
ity, and diagnostic utility. Paper presented at the annual
meeting of the International Society for Traumatic Stress
Studies, San Antonio, TX.
Wells, A. (2000). Emotional disorders and metacognition:
Innovative cognitive therapy. Chichester, England: Wiley.
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from

10 Takarangi et al.
Wells, A., & Cartwright-Hatton, S. (2004). A short form of the
metacognitions questionnaire: Properties of the MCQ-30.
Behaviour Research Therapy, 42, 385–396. doi:10.1016/
S0005-7967(03)00147-5
Wells, A., & Sembi, S. (2004). Metacognitive therapy for PTSD: A
preliminary investigation of a new brief treatment. Journal
of Behaviour Therapy and Experimental Psychiatry, 35,
307–318. doi:10.1016/j.jbtep.2004.07.001
Woud, M. L., Holmes, E. A., Postma, P., Dalgleish, T.,
&Mackintosh, B. (2012). Ameliorating intrusive memories
of distressing experiences using computerized reappraisal
training. Emotion, 12, 778–784. doi:10.1037/a0024992
Zigmund, A. S., & Snaith, R. P. (1983). The hospital anxiety
and depression scale. Acta Psychiatrica Scandinavica, 67,
361–370. doi:10.1111/j.1600-0447.1983.tb09716.x
at JOHN JAY COLLEGE on October 3, 2016cpx.sagepub.comDownloaded from