1 The Mediating role of Cognitive Biases in the - UvA-DARE

The Mediating role of Cognitive Biases in the Relationship between Anxiety and Paranoia
Anika Vermeulen
10357009
Universiteit van Amsterdam
Clinical Psychology
L.L.N.J. Boyette
4.580 words
1
Abstract
Background. There is evidence that paranoia forms a continuum, making it possible to draw
conclusions about the development of clinical paranoia by studying non-clinical populations.
Freeman et al.’s threat anticipation model holds that multiple factors, including affective and
cognitive components, produce paranoia. Anxiety may trigger biased interpretation (cognitive
biases), giving rise to paranoia.
Aims. Our first aim was to test which cognitive biases were related to and predicted paranoia
in a non-clinical population. Our second aim was to test whether these cognitive biases
mediated the relationship between anxiety and paranoia.
Method. In this cross-sectional study, a sample of 192 individuals from the general
population were administered three self-report questionnaires measuring anxiety, cognitive
biases and paranoia. Correlations and a stepwise regression assessed which cognitive biases
were associated with and predicted paranoia. To test the indirect relationship between anxiety
and paranoia via cognitive biases, a mediation analysis was conducted.
Results. Significant correlations were found between six reasoning biases and paranoia.
Three reasoning biases, attention for threat, external attribution bias and social cognition
problems predicted paranoia most strongly. Anxiety significantly predicted paranoia, and
these three reasoning biases were found to mediate this relationship.
Conclusions. The results corroborate the threat anticipation model. Longitudinal research is
needed to draw causal conclusions about the relationship between anxiety, reasoning biases
and paranoia.
2
Introduction
People with paranoia ascribe harmful intentions to others, which they see as aimed at
themselves. This paranoid interpretation may be adaptive in some situations, becoming only
clinically relevant when the paranoid thoughts are “excessive, exaggerated, unfounded and
distressing” (Freeman et al., 2005), and, in the most severe form, can reach delusional
intensity when they are resistant to change, even when challenged by conflicting evidence
(Oltmanns, 1988, cited in Freeman, 2007).
There is evidence that attenuated forms of paranoia are quite frequent in the general
population. For example, Freeman et al. (2005) found that between 30 and 40% of a student
sample experienced weekly mild suspicions, while thoughts increasingly persecutory in
nature were less common and were associated with the highest levels of distress. Odder
thoughts (e.g. “There is a possibility of a conspiracy against me”) were less frequently
endorsed than more common ones (e.g. “There might be negative comments being circled
about me”). The rarer the thought, the higher the reported level of distress. This finding shows
that degrees of paranoia are present in the general population. These non-clinical experiences
seem to be related to the same risk factors as those for clinical paranoia, such as “urban
dwelling, living alone, depression” (Freeman, 2007). This supports the conceptualization of
paranoia as a continuum (Myin-Germeys, Krabbendam & van Os, 2003, cited in Freeman,
2006) with non-clinical paranoid thoughts at one end, and full-blown, distressing paranoia at
the other. Therefore, studying paranoia in the general population may shed light on the
developmental process of clinical paranoia. Finding its origins may also aid in the
development of specified therapy forms of treatment.
According to Freeman and colleagues’ (e.g. Freeman & Garety, 2004, cited in
Freeman, 2007) threat anticipation model, paranoia is the result of multiple factors. Threat
beliefs, reasoning processes and (emotional) arousal, most importantly anxiety, are given a
3
prominent role in the development of paranoia (Freeman, 2007). Biased reasoning processes
are thought to lead the individual to suspicious, unfounded and rigid interpretation. Anxiety
steers the individual towards feeling threatened and persecuted. At higher levels, anxiety
limits the use of thorough information processing in favor of heuristic thinking (Lincoln et al.,
2010). Accompanied by such cognitive shortcuts, anxious thoughts can intensify and
approach paranoid, delusional proportions. Cognitive biases may also maintain the delusional
process, restricting access to threat-disconfirming evidence. In other words, anxiety and
appraisal may predict paranoia.
Distinctive reasoning processes have been found to be associated with paranoia.
Firstly, people with paranoid ideation require little evidence on which they base their
conclusions (Huq, Garety & Hemsley, 1988). This way of thinking defines a jumping-toconclusions bias (Merrin, 2007). This bias was found more frequently in clinical than nonclinical paranoia (Merrin, 2007) and is associated with higher degrees of delusional
conviction (Freeman et al., 2008). A second noteworthy cognitive bias is the external
attribution bias. People with this bias see others or situational factors as the cause of positive
or negative events (Craig et al., 2004). In the case of paranoia, the sufferer feels threatened
and attributes this to the intentions of others, making an external attribution style for negative
events likely. Research has indeed indicated that people with paranoia tend to make external
attributions for negative events (Craig et al., 2004; Fear, Sharp & Healy, 1996; Krstev et al.,
1999). Attention for threat is a third relevant bias. The anticipation of harm (Freeman, 2007)
may make a person more vigilant to signs of danger, requiring preferential attention to threat
stimuli. Studies using the Emotional Stroop Task (Stroop, 1935, cited in Lim, Gleeson &
Jackson, 2011) have found some evidence for this bias in people with persecutory delusions
(Fear, Sharp & Healy, 1996; Bentall & Kaney 1994; Lim, Gleeson & Jackson, 2011). A
fourth cognitive bias in persecutory delusions is belief inflexibility. This resistance of
4
alternatives would by definition be at play in delusional disorders, as these are characterized
by unamenable beliefs. Garety et al. (2005) found that currently deluded patients applied this
bias, while Savulich et al. (2015) found that it predicted paranoid interpretation. A final
cognitive bias is safety behavior. Anticipation of harm may drive people with paranoia to seek
safety (Freeman & Garety, 1999; Freeman et al., 2001, cited in Freeman, Garety & Kuipers,
2001). Freeman, Garety and Kuipers (2001) hypothesized that these safety measures prevent
sufferers of persecutory delusions from being confronted with delusion-contradictory
information. Also, they are prevented from experiencing the threat not occurring without
having used a safety behavior. To the paranoid person, the absence of danger does not
disprove the delusion, it confirms the success of the safety measure (Salkovski, 1991, cited in
Freeman, Garety & Kuipers, 2001). This could increase the chances of safety behaviors being
used in the future, maintaining paranoid ideation. Based on this reasoning, the authors
administered the Safety Behaviours Questionnaire - Persecutory Beliefs to a group of people
with clinical paranoia. All participants had used at least one safety behavior in the past month.
Higher levels of anxiety were associated with using more safety behaviors, avoidance being
the most common. Generally, participants judged these measures as having been successful.
Research also supports the assumption of Freeman et al.’s (Freeman, 2007) threat
anticipation model that anxiety and paranoia are related. Studies have found associations
between anxiety and clinical as well as non-clinical paranoid ideation (Martin & Penn, 2001).
Anxiety contributes to and predicts paranoia (Freeman et al., 2003; 2005; 2005b; 2012).
Paranoid people have been found to experience severe anxiety (Startup et al., 2007), in
frequencies comparable to those in GAD (Garety & Freeman, 1999). In summary, cognitive
biases and anxiety are both related to paranoia. However, the threat anticipation model
(Freeman, 2007) also poses that anxiety creates conditions under which biased reasoning
becomes more likely. If this is the case, anxiety may give rise to cognitive biases, which in
5
turn affect paranoid ideation. In other words, cognitive biases may mediate the relationship
between anxiety and paranoia. Previous research has found such an effect for jumping to
conclusions (Lincoln et al. 2010; Galbraith et al., 2014) and safety behavior (Freeman, Garety
& Kuipers, 2001), but other cognitive biases have not been studied as such.
Therefore, the first aim of the current study was to identify the cognitive biases related
to paranoia in a non-clinical population. Previous research on non-clinical populations mostly
studied associations between jumping to conclusions, external attribution bias, attention for
threat and paranoia. The current study set out to expand this with other relevant biases, such
as cognitive inflexibility and safety behavior. The second, and main, objective was to
determine whether these cognitive biases mediate the relationship between anxiety and
paranoia. The hypotheses for this study were as follows: jumping to conclusions, external
attribution bias and attention for threat bias would significantly correlate at least r=.3 with
paranoia. These biases have strong evidence supporting their presence in paranoia and have
also been found to be applied by members of the general population (Merrin, 2007; Fear,
Sharp & Healy, 1996; Lim, Gleeson & Jackson, 2011). Yung et al. (2005, cited in Bastiaens et
al., 2013) argue that safety behavior is likely characteristic of the clinical population only, as
this bias constitutes a behavioral change in response to perceived threat. However, this bias
has not been studied in the general population, so it is unclear whether Yung et al.’s reasoning
holds. Therefore, we will perform an exploratory assessment of this bias in relation to
paranoia. The same will be done for cognitive inflexibility bias, as this bias seems
characteristic of delusional disorders and may not be present in the general population. In
regard to the second aim, anxiety is expected to significantly predict the total paranoia score.
Based on the literature, anxiety is also expected to significantly predict three cognitive biases:
jumping to conclusions, attention for threat and safety behavior (Lincoln et al., 2010;
MacLeod & Mathews, 1991; Freeman, Garety & Kuipers, 2001). The relationship between
6
anxiety and paranoia will weaken when these cognitive biases are added as mediators. To our
knowledge, there is no research on the relationship between anxiety and external attribution
bias, belief inflexibility and social and subjective cognition problems. Therefore, we will
conduct explorative analyses of these relationships.
Methods
Participants
All participants signed an informed consent document before taking the survey. Subjects were
university students and members of the researchers’ social circles. They were recruited
through advertising the study on social media and personal communication. Participants were
also recruited through the online system (DPMS) used by the University of Amsterdam to
advertise experiments. Psychology students are required to take part in such experiments and
gain course credits for doing so. The sample consisted of slightly more psychology students
(52%) than personally recruited participants (48%). Students of the University of Amsterdam
were awarded one research credit for their participation. The current study was part of a larger
test battery, which took approximately 1 hour to complete.
Measures
All questionnaires were self-report measures, presented as online surveys. Through access
with a web-based link, participants anonymously completed the test battery on their own
computer, smartphone or tablet.
Anxiety
Anxiety was assessed using the anxiety subscale of the Dutch version of the Depression,
Anxiety and Stress Scale (DASS, S. H. Lovibond & P. F. Lovibond, 1995). This self-report
7
questionnaire consists of 42 items describing experiences indicating stress, depression or
anxiety. Based on a sample of both clinical and non-clinical participants, all three scales
(depression, anxiety and stress) were found to have high reliability (Cronbach’s alpha =.97,
.92 and .95, respectively) and validity (Antony et al. 1998). In the current study, Cronbach’s
alpha for the anxiety subscale was .89. The participant is instructed to report whether, in the
past week, an experience occurred never (1 point), sometimes (2 points), often (3 points) or
most of the time (4 points). Higher scores indicate higher levels of the corresponding affective
state. The total score on the anxiety subscale is the sum of 14 anxiety items, such as “I felt
scared without any good reason” (Antony et al., 1988).
Cognitive biases
The Davos Assessment of Cognitive Biases Scale (DACOBS, Van der Gaag et al., 2013) was
used to measure cognitive biases. It consists of 42 items indicative of seven subscales:
jumping to conclusions, belief inflexibility, attention for threat, external attribution bias,
social cognition problems, subjective cognition problems and safety behavior. Based on a
sample consisting of non-clinical as well as clinical participants, Van der Gaag et al. (2013)
reported Cronbach’s alphas for each of these scales: .72, .74, .71, .64, .76, .69 and .82
respectively. In our study, we found alpha values of .61, .77, .74, .74, .75, .76 and .93.
Two biases measured by this scale were not mentioned before. Social cognition
problems are comparable to Theory of Mind issues, in which the person has difficulty taking
another’s perspective. Subjective cognition problems concern verbal intelligence (Van der
Gaag et al., 2013). The results of a new literature search indicated that social cognition
problems had been found in people with paranoia (Freeman, 2007). Therefore, we also
assessed social cognition problems in the further analyses. To our knowledge, there is no
literature on the relationship between subjective cognition problems and paranoia. To assess
8
this ourselves, we decided to do exploratory analyses of this bias in relation to anxiety and
paranoia. All DACOBS subscales except social and subjective cognition were successfully
validated using a non-clinical and clinical sample. For this reason, any conclusions about
these two biases would be drawn with caution.
An example of an item reflecting safety behavior is “I don’t go out after dark”. “I’m
on the lookout for danger” is indicative of attention for threat. Per item, the participant
reports to what degree he or she agrees or disagrees with the statement using a 7-point Likert
scale: completely agree (7 points), agree (6 points), slightly agree (5 points), neither agree nor
disagree (4 points), slightly disagree (3 points), disagree (2 points), completely disagree (1
point). Certain items correspond to these cognitive biases and the summed scores of these
items reflect the participant’s score on each subscale. These scores can be compared to norm
scores for schizophrenia spectrum patients (N=138) and normal control participants (N=186),
collected by the DACOBS’ authors.
Paranoia
Paranoia was assessed using the Green Paranoid Thoughts Scales (GPTS, Green et al., 2008).
This self-report questionnaire consists of two 16-item scales, the first assessing ideas of social
reference and the second ideas of persecution. Every item is a statement reflecting either
topic, such as “I often heard people referring to me” and “I was upset about being followed”,
respectively. Using a five-point Likert scale, the participant judges to what extent he or she
experienced the described situation in the last month. Participants responded with “not at all”
(1 point), “somewhat” (3 points), “totally” (5 points) or with two unlabeled response options
in between these answers (2 points; 4 points). The total score is the sum of the scores on
every item. Higher scores indicate higher degrees of paranoia. Based on a clinical and nonclinical sample, the total GPTS was validated and was found to have high reliability,
9
Cronbach’s alpha = .95 (non-clinical sample, Green et al., 2008). Our study found a
comparable alpha value of .96.
Analyses
All analyses were conducted using IBM SPSS 20. The first step in the analyses was to
assess the normality of the scores on the subscales of the DACOBS (cognitive biases), the
GPTS (paranoid ideation) and the DASS anxiety subscale. Because the assessed population
was a non-clinical one, a positively skewed distribution was expected for all scales.
Therefore, normality was tested using the Kolmogorov-Smirnov test and by visual inspection
of histograms. If the scales were positively skewed, we would correct for this using nonparametric tests, when possible. Also, bootstrapping was applied. Next, we would assess
which cognitive biases were related to paranoia, by use of Pearson’s or Spearman’s
correlations. Higher, positive correlations indicated that the measured cognitive bias went
along with paranoia. Based on the literature reviewed, we expected all DACOBS subscales,
possibly except subjective cognition problems, to be positively correlated with paranoia.
Biases found to be correlated at least r=.3 (medium effect, Cohen, 1988, cited in Hemphill,
2003) with paranoia would be included in the further analysis, if they did not violate the
assumption of multicollinearity. If correlations between cognitive biases were higher than
r=.8 (Cohen, 1988, cited in Hemphill, 2003), these biases could not be seen as independent
predictors. Biases significantly correlated at least r=.3 with the total GPTS score would be
entered into a stepwise regression, which would show which biases strongly predict paranoia
(provided unique contribution). Regarding the second aim of this study, all DACOBS biases
were entered into a mediation analysis using the INDIRECT script by Andrew F. Hayes for
SPSS (Preacher & Hayes, 2008). This analysis tested a mediation model, in which cognitive
biases mediate the relationship between anxiety and paranoia. The first step in the mediation
10
analysis was to assess whether the predictor, anxiety, significantly predicted paranoid
ideation. If so, a further mediation analysis would be justified. The mediation analysis
assessed 1) whether anxiety significantly predicted cognitive biases, 2) whether cognitive
biases significantly predicted paranoia and 3) whether the relationship between anxiety and
GPTS weakened when cognitive biases were entered as mediators. If these statistics were
significant, a mediation effect could be said to be present.
Results
Demographics
The initial data set consisted of 256 participants. Data from 64 participants was
eliminated due to incomplete or deemed unreliable reporting, indicated as an unfinished
report (N=55) and surveys that were taken in less than 20 minutes (N=9). The surveys taken
in less than 20 minutes were deemed unreliable because they were unlikely to have been
completed seriously. As such, they probably did not reflect the participant’s true attitude,
opinion or experience. The final data sample consisted of 192 participants, of which 76%
(N=146) were female and 24% (N=46) were male. The mean age was 26.71 (sd = 11.23) with
a minimum of 18.5 and a maximum of 66.1 years old.
All scales except the DACOBS attention for threat subscale were positively skewed.
Psychology students represented 52% (N=99) of the sample, while personally recruited
participants represented the other 48% (N=93). Psychology students were significantly
younger than personally recruited participants, p<.001. A chi-square test indicated that the
ratio of men to women was different for the Psychology student group relative to the
personally recruited group, 𝜒𝜒 2 (1) = 6.82, p<.01. There were fewer men but more women in
the former than in the latter group. Psychology students and personally recruited participants
scored similarly on all scales except for the DACOBS jumping to conclusions, belief
11
inflexibility and attention for threat subscales. A Mann-Whitney independent samples t-test
showed that Psychology students scored significantly lower than personally recruited
participants on these three measures. This was not explained by an effect for age, which was
entered as a covariate in bootstrapped regressions, with participant status (psychology student
or not) as the predictor variable and DACOBS jumping to conclusions, belief inflexibility and
attention for threat as outcome variables (p=.41, p=.78 and p=.81, respectively). Gender was
not related to any of the scales. Age, which was correlated r=-.214, p<.01 with the total GPTS
score, was entered as a covariate in the further analyses.
Descriptives
On average, 12.12% of our sample reported having paranoid thoughts more often than
never in the past month. The scores on the GPTS items were mostly low, ranging between 1
and 2 points. The highest mean score (2.22) was reported for GPTS item 7, “I believed that
certain people were not what they seemed” (Green et al., 2008). Descriptive statistics for the
DASS anxiety subscale, GPTS and DACOBS subscales are shown in Table 1. The results of
the current study were compared with norm scores for the DACOBS subscales (van der Gaag
et al., 2013), shown in the subscript under Table 1. The mean scores on the DACOBS
subscales reported by Van der Gaag et al. (2013) were calculated using the results of clinical
as well as non-clinical participants. As could be expected, these means were higher than those
found in the current, non-clinical sample.
12
Table 1. Descriptive statistics of measured variables.
Mean
Median
SD
Minimum to Maximum
Classification
DASS anxiety
19.15
17
5.86
14-47
GPTS
46.10
41
17.13
32-147
DACOBS
24.75
25
4.97
11-37
Average
DACOBS Belief inflexibility
15.70
15
5.63
6-38
Above average
DACOBS Attention for threat
19.90
20
6.57
6-37
Average
DACOBS External attribution
16.39
16
5.83
6-40
Average
DACOBS Social cognition
19.06
18
6.03
6-35
Above average
DACOBS Subjective cognition
19.82
20
6.39
6-37
Above average
DACOBS Safety behavior
10.75
8
7.42
6-42
Above average
For comparison purposes: Van der Gaag et al. (2013) means and classification. Jumping to conclusions: 25.59 (Average), Belief inflexibility:
20.72 (High), Attention for threat: 25.75 (High), External attribution: 22.93 (High), Social cognition: 24.28 (High), Subjective cognition: 24.05
(High), Safety behavior: 16.06 (Above average).
13
Table 2. Correlations among DACOBS subscales and between DACOBS subscales and GPTS total score.
GPTS
Jumping to
Belief
Attention for
External
conclusions
inflexibility
threat
attribution
Social cognition
Subjective cognition
Safety behavior
GPTS
-
0.13
0.35***
0.47***
0.40***
0.50***
0.28***
0.36***
Jumping to
0.13
-
0.34***
0.23***
0.24***
0.10
0.00
0.70
conclusions
Belief inflexibility
0.35*** 0.34***
-
0.46***
0.58***
0.50***
0.48***
0.51***
Attention for threat
0.47*** 0.23***
0.46***
-
0.55***
0.51***
0.41***
0.48***
External attribution
0.40*** 0.24***
0.58***
0.55***
-
0.53***
0.33***
0.51***
Subjective cognition
0.50*** 0.10
0.50***
0.51***
0.53***
-
0.57***
0.40***
Social cognition
0.50**
0.00
0.48***
0.41***
0.33***
0.57***
-
0.30***
Safety behavior
.36***
0.07
0.51***
0.48***
0.51***
0.40***
0.30***
-
**p<.01, *** p<0.001.
14
Correlations
The next step was to calculate the correlations among the DACOBS subscales and
between these subscales and the total GPTS score. Because the majority of the variables did
not follow a normal distribution, Spearman correlation coefficients were calculated instead of
Pearson correlation coefficients. The results can be seen in Table 2. The correlations between
the subscales did not exceed r =.6, so no serious intercorrelations (r >.8) were present. The
correlations between the total GPTS score and all DACOBS subscales were positive. Only the
correlations between JTC and GPTS were lower than r=.3. All other correlations were
significant. Belief inflexibility, attention for threat, external attribution bias, social cognition,
subjective cognition and safety behavior were correlated with at least moderate strength (r >
0.3) with the total GPTS score.
Regression analysis
Two blocks of stepwise regressions were conducted, one with age as the predictor
variable and the other with the six DACOBS subscales that significantly correlated with
paranoia at least r=.3 as predictor variables. This showed that the biases that contributed the
most variance to the total paranoia score were external attribution bias, b =1.29, se = 0.24,
95% BCI [0.81-1.78], p< 0.001, attention for threat, b =0.75, se = 0.20, 95% BCI [0.371.14], p<.001 and social cognition problems b =0.87, se = 0.22, 95% BCI [0.44-1.29], p<.05,
after controlling for age. The total model explained about 43% of the variance in the total
GPTS score (𝑅𝑅 2 = .43, F=32.92, p<.001).
Mediation
Using Andrew Hayes’ INDIRECT script for IBM SPSS, all DACOBS subscales were
entered as potential mediating variables in the relationship between anxiety and paranoia. The
15
results are shown in Fig. 1. It was found that the DASS anxiety subscale significantly
predicted the total GPTS score, b=1.87, se=.17, t= 11.02, p<.001. Therefore, a further
mediation analysis was justified. Attention for threat, external attribution bias and social
cognition problems were found to mediate the relationship between anxiety and paranoia: 1)
Anxiety significantly predicted attention for threat, b= .32 se=.076 t= 4.15 p<.001, external
attribution bias, b= .31, se= .06, t= 5.10, p<.001 and social cognition b= .36, se=.07, t= 5.35,
p<.001 2) These three biases significantly predicted the total GPTS score, b=.38, se=.18,
t=2.05, p<.05; b=.62, se=.24, t=2.58, p<.05; b=.79, se=.21, t=2.58, p<.05, respectively.
Conditions 1 and 2 were significant, indicating unique mediation effects on the relationship
between anxiety and paranoia. After bootstrapping, this was confirmed (bias corrected 95%
CI total: 0.18-0.81, attention for threat 0.02-0.28, external attribution bias 0.02-0.47, social
cognition problems 0.13-0.52). 3) When attention for threat, external attribution bias and
social cognition problems were added to the regression, anxiety still significantly predicted
the total GPTS score, but less strongly so b=1.39, se=.17, t=11.02, p<.01. The full model
accounted for approximately 57% of the total variance in paranoia (𝑅𝑅 2 =.57, F=29.21,
p<.001).
16
Figure 1. Mediation of cognitive biases in the relationship between anxiety and paranoia. *p<.05, **<.001
17
Discussion
All cognitive biases except JTC (r=.13) and social cognition (r=.28) correlated at least
with moderate strength with the total paranoia score. A stepwise regression showed that
attention for threat, external attribution bias and social cognition problems were the strongest
predictors of paranoia. These biases were found to partially mediate the relationship between
anxiety and paranoia. As the threat anticipation model (Freeman et al., 2007) predicted,
cognitive biases mediated the relationship between anxiety and paranoia. These relationships
seem plausible. An anxious state may drive an individual to be on the lookout for signals of
danger, requiring preferential attention to threat-related stimuli. The individual may become
overly sensitive to such information and see harmful intent where there is none, such as in
ambiguous facial expressions (Freeman, 2007), resulting in paranoia. A similar pathway
might form through an external attribution bias. Higher levels of anxiety increase the chances
of using heuristic judgment (Lincoln, 2010). Blaming others for negative events may be easier
than looking for other causal factors. The sufferer may become quick to see others as the
cause of adversities, paving the way for paranoia. The mediating role of social cognition
problems is less convincing. Firstly, the DACOBS subscale for this bias was not validated
(Van der Gaag et al., 2013). Secondly, there is little evidence supportive of the idea that social
cognition problems predict paranoia. This bias, more commonly known as Theory of Mind
problems (Freeman, 2007) has not consistently been found in people with paranoia (e.g.
Walston et al., 2000, cited in Freeman, 2007) and may be caused by other symptoms of
present mental disorders, such as negative symptoms and thought disorder. Considering these
reasons, it is unlikely that our findings on this bias are reliable. Future studies should use a
validated measure of social cognition problems, such as the false belief task (Wimmer &
Perner, 1985).
In the current study, around 12% of our non-clinical sample reported having had
18
paranoid thoughts at least ‘sometimes’. However, in Freeman et al.’s (2005) student sample,
30-40% reported paranoid thoughts. This may have been due to the instruments used to
measure paranoia. Freeman and colleauges used the Paranoia Checklist, which, unlike the
DACOBS, was designed specifically to measure paranoia in college students (Freeman et al.,
2005). This may have made the statements in the questionnaire more applicable to students
than the DACOBS’ statements, resulting in overreporting in Freeman et al.’s (2005) study, or
underreporting in our own.
A surprising finding was that the correlation between JTC and paranoia was quite
weak (r=.19) in the current sample. This bias has been researched in relation to paranoia and
the evidence for this association is substantial (Freeman, 2007). Therefore, we expected to
find similar results in the current study. The reliability of the DACOBS jumping-toconclusions subscale (Cronbach’s alpha =.61) was below the ‘acceptable’ score of .7 (Field,
2009), while Van der Gaag et al. (2013) found an alpha value of .72. This may explain the
absence of an association between jumping to conclusions and paranoia in the current study,
but another reason may be that the studies described in Freeman’s (2007) review used a reallife task to measure this bias. This bead task is a probabilistic reasoning task in which
participants “decide from which of two hidden jars coloured beads are being drawn. The jars
both contain beads of two different colours but the proportion of beads of each colour in the
jars is reversed.” (Freeman, 2007). People who jump to conclusions require seeing fewer
beads before drawing their conclusion than non-clinical controls (Freeman, 2007). The
DACOBS jumping to conclusions subscale may not have been comparable to the decision
making task. People may be more inclined to judge themselves as informed decision makers,
so when presented with a question such as “I don't need to evaluate all the facts to reach a
conclusion”, they may try to convey this image. The bead task may be a more objective way
of testing this bias, as the person is required to draw an actual conclusion without necessarily
19
knowing what the experiment is measuring.
As the threat anticipation model (Freeman, 2007) poses, paranoia may result from
several factors. One relevant factor not considered in the current study was stress. Stressful
life events may precipitate paranoid ideation, often in the context of a history of anxiety
(Freeman et al., 2002). The arousal caused by stress triggers a search for meaning in people
vulnerable to paranoia. Cognitive biases may then be applied, leading to paranoid
interpretation. This implicates that there may be an overlooked role for stress in the current
study. Trait anxiety may form the circumstances under which a person becomes vulnerable to
stress. In the event of trauma, isolation or other adverse occurrences, the person resorts to
quick-fix thinking to find meaning in this heightened state of arousal. As discussed in the
current study, biased reasoning may then lead to paranoid interpretation. Longitudinal
research can test this model, measuring the effect of trait anxiety on stress, stress on the use of
cognitive biases and the combined effect of these factors on the development of paranoia.
A limitation of this study was its use of online self-report surveys, which may have
affected the data. As participants were in an uncontrolled, unmonitored environment, the
presence of confounders such as distractions may have been more likely. Anoter limitation
may have been that most participants (52%) were psychology students that may have been
familiar with the test battery and its purpose. This may have lead participants to give socially
desirable answers to avoid scoring highly on the underlying construct. Indeed it was found
that psychology students scored significantly lower than non-psychology students on JTC,
belief inflexibility bias and attention for threat. Using a similar questionnaire with a validation
scale to detect any biased response patterns could minimalize this. A third limitation is the
cross-sectional nature of our study. This prevents us from drawing causal conclusions about
the pathway to paranoia. However, Lincoln et al. (2010) found that non-clinical participants in
an induced anxiety condition jumped to conclusions more than did participants in a neutral
20
condition. These anxious participants also scored higher on a paranoia scale. Therefore, an
interesting area of future study may be to conduct similar experimental studies assessing other
cognitive biases as mediators in the relationship between induced anxiety and paranoia. The
results of the current study indicate that similar results may be expected for attention for
threat, external attribution bias and social cognition problems.
Concluding, this study showed that multiple factors contribute to paranoia, supporting
the threat anticipation model (Freeman, 2007). The findings suggest that treatment should
target anxiety and reasoning processes, such as through relaxation and metacognitive therapy.
21
References
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental
disorders: DSM-5. Washington, D.C: American Psychiatric Association.
Antony, M. M., Bieling, P. J., Cox, B. J., Enns, M. W., & Swinson, R. P. (1998).
Psychometric properties of the 42-item and 21-item versions of the Depression
Anxiety Stress Scales in clinical groups and a community sample. Psychological
assessment, 10(2), 176.
Bastiaens, T., Claes, L., Smits, D., De Wachter, D., van der Gaag, M., & De Hert, M. (2013).
The Cognitive Biases Questionnaire for Psychosis (CBQ-P) and the Davos
Assessment of Cognitive Biases (DACOBS): validation in a Flemish sample of
psychotic patients and healthy controls. Schizophrenia research, 147(2), 310-314.
Bentall, R. P., Kinderman, P., & Kaney, S. (1994). The self, attributional processes and
abnormal beliefs: towards a model of persecutory delusions. Behaviour research and
therapy, 32(3), 331-341.
Craig, J. S., Hatton, C., Craig, F. B., & Bentall, R. P. (2004). Persecutory beliefs, attributions
and theory of mind: comparison of patients with paranoid delusions, Asperger's
syndrome and healthy controls. Schizophrenia research,69(1), 29-33.
Fear, C., Sharp, H., & Healy, D. (1996). Cognitive processes in delusional disorders. The
British Journal of Psychiatry, 168(1), 61-67.
Field, A. (2013). Discovering statistics using IBM SPSS statistics. Sage. Freeman, D. (2007).
Suspicious minds: the psychology of persecutory delusions. Clinical Psychology
review, 27(4), 425-457.
Freeman, D., & Garety, P. A. (1999). Worry, worry processes and dimensions of delusions: an
exploratory investigation of a role for anxiety processes in the maintenance of
delusional distress. Behavioural and Cognitive Psychotherapy,27(01), 47-62.
22
Freeman, D., Dunn, G., Garety, P. A., Bebbington, P., Slater, M., Kuipers, E., ... & Ray, K.
(2005). The psychology of persecutory ideation I: a questionnaire survey. The
Journal of nervous and mental disease, 193(5), 302-308.
Freeman, D., Garety, P. A., & Kuipers, E. (2001). Persecutory delusions: developing the
understanding of belief maintenance and emotional distress.Psychological
medicine, 31(07), 1293-1306.
Freeman, D., Garety, P. A., Kuipers, E., Fowler, D., & Bebbington, P. E. (2002). A cognitive
model of persecutory delusions. British Journal of Clinical Psychology, 41(4), 331347
Freeman, D., Garety, P. A., Bebbington, P. E., Smith, B., Rollinson, R., Fowler, D., &
Dunn, G. (2005). Psychological investigation of the structure of paranoia in a non
clinical population. The British Journal of Psychiatry, 186(5), 427-435.
Freeman, D., Garety, P. A., Bebbington, P., Slater, M., Kuipers, E., Fowler, D., ... & Dunn,
G. (2005). The psychology of persecutory ideation II: a virtual reality experimental
study. The Journal of nervous and mental disease, 193(5), 309-315.
Freeman, D., Gittins, M., Pugh, K., Antley, A., Slater, M., & Dunn, G. (2008). What
makes one person paranoid and another person anxious? The differential prediction
of social anxiety and persecutory ideation in an experimental situation. Psychological
medicine, 38(08), 1121-1132.
Freeman, D., Pugh, K., & Garety, P. (2008). Jumping to conclusions and paranoid
ideation in the general population. Schizophrenia research, 102(1), 254-260.
Freeman, D., Slater, M., Bebbington, P. E., Garety, P. A., Kuipers, E., Fowler, D., &
Vinayagamoorthy, V. (2003). Can virtual reality be used to investigate persecutory
ideation?. The Journal of nervous and mental disease, 191(8), 509-514.
23
Freeman, D., Stahl, D., McManus, S., Meltzer, H.,Brugha, T., Wiles, N., & Bebbington, P.
(2012). Insomnia, worry, anxiety and depression as predictors of the occurrence and
persistence of paranoid thinking. Social psychiatry and psychiatric
epidemiology, 47(8), 1195-1203.
Galbraith, N. D., Manktelow, K. I., Chen-Wilson, C. H., Harris, R. A., & Nevill, A. (2014).
Different Combinations of Perceptual, Emotional, and Cognitive Factors Predict Three
Different Types of Delusional Ideation During Adolescence. The Journal of nervous
and mental disease, 202(9), 668-676.
Garety, P. A., Freeman, D., Jolley, S., Dunn, G., Bebbington, P. E., Fowler, D. G., ... &
Dudley, R. (2005). Reasoning, emotions, and delusional conviction in
psychosis. Journal of abnormal psychology, 114(3), 373.
Green, C. E. L., Freeman, D., Kuipers, E., Bebbington, P., Fowler, D., Dunn, G.,& Garety, P.
A. (2008). Measuring ideas of persecution and social reference: the Green et al.
Paranoid Thought Scales (GPTS). Psychological medicine, 38(01), 101-111.
Hemphill, J. F. (2003). Interpreting the magnitudes of correlation coefficients.
Huq, S. F., Garety, P. A., & Hemsley, D. R. (1988). Probabilistic judgements in deluded and
non-deluded subjects. The Quarterly Journal of Experimental Psychology, 40(4), 801812.
Krstev, H., Jackson, H., & Maude, D. (1999). An investigation of attributional style in first
episode psychosis. British Journal of Clinical Psychology, 38(2), 181-194.
Lim, M. H., Gleeson, J. F., & Jackson, H. J. (2011). Selective attention to threat bias in
delusion-prone individuals. The Journal of nervous and mental disease, 199(10), 765772.
24
Lincoln, T. M., Lange, J., Burau, J., Exner, C., & Moritz, S. (2010). The effect of state
anxiety on paranoid ideation and jumping to conclusions. An experimental
investigation. Schizophrenia Bulletin, 36(6), 1140-1148.
MacLeod, C., & Mathews, A. (1991). Biased cognitive operations in anxiety: accessibility of
information or assignment of processing priorities?. Behaviour research and
therapy, 29(6), 599-610.
Martin, J. A., & Penn, D. L. (2001). Social cognition and subclinical paranoid
ideation. British Journal of Clinical Psychology, 40(3), 261-265.
McLean, C. P., & Anderson, E. R. (2009). Brave men and timid women? A review of the
gender differences in fear and anxiety. Clinical Psychology review,29(6), 496-505.
Merrin, J., Kinderman, P., & Bentall, R. P. (2007). ‘Jumping to conclusions’ and attributional
style in persecutory delusions. Cognitive Therapy and Research,31(6), 741-758.
Perner, J., & Wimmer, H. (1985). “John thinks that Mary thinks that…” attribution of second
order beliefs by 5-to 10-year-old children. Journal of experimental child
Psychology, 39(3), 437-471.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing
and comparing indirect effects in multiple mediator models. Behavior Research
Methods, 40, 879-891
Savulich, G., Freeman, D., Shergill, S., & Yiend, J. (2015). Interpretation Biases in
Paranoia. Behavior therapy, 46(1), 110-124.
Startup, H., Freeman, D., & Garety, P. A. (2007). Persecutory delusions and catastrophic
worry in psychosis: developing the understanding of delusion distress and
persistence. Behaviour research and therapy, 45(3), 523-537.
25
Van der Gaag, M., Schütz, C., Ten Napel, A., Landa, Y., Delespaul, P., Bak, M., & De Hert,
M. (2013). Development of the Davos assessment of cognitive biases scale
(DACOBS). Schizophrenia research, 144(1), 63-71.
26