Frederick Aardema1, Kieron P. O'Connor1 and Paul M. G. Emmelkamp2. 1Centre de .... Reed, 1977), and hence valid deductive con- clusions may be drawn ...
Cognitive Behaviour Therapy Vol 35, No 3, pp. 138–147, 2006
Inferential Confusion and Obsessive Beliefs in Obsessive-Compulsive Disorder Frederick Aardema1, Kieron P. O’Connor1 and Paul M. G. Emmelkamp2 1
Centre de Recherche Fernand-Seguin, Montre´al, Que´bec, Canada, and 2University of Amsterdam, Department of Clinical Psychology, Amsterdam, The Netherlands
Abstract. The goal of the present study was to investigate whether inferential confusion could account for the relationships between obsessional beliefs and obsessive-compulsive disorder (OCD). The Inferential Confusion Questionnaire and the Obsessive Beliefs Questionnaire were administered to a sample of 85 participants diagnosed with OCD. Results showed that the relationship between obsessive beliefs and obsessive-compulsive symptoms decreased considerably when controlling for inferential confusion. Conversely, the relationship between inferential confusion and obsessivecompulsive symptoms was not substantially affected when controlling for obsessive beliefs. Since inferential confusion has an overlap with overestimation of threat, a competing hypothesis for the results was investigated. Results indicated that inferential confusion was factorially distinct from overestimation of threat, and that the independent construct of inferential confusion remains significantly related to obsessive-compulsive symptoms when controlling for anxious mood. These results are consistent with the claim that inferential confusion may be a more critical factor in accounting for OCD symptoms than are obsessive beliefs and appraisals. Key words: obsessivecompulsive disorder; inference-based approach; inferential confusion; reasoning; beliefs. Received August 3, 2005; Accepted February 2, 2006 Correspondence address: Frederick Aardema, Centre de Recherche Fernand-Seguin, 7331 Hochelaga, Montre´al, Que´bec, H1N 3V2, Canada. Tel: +514-251-4015; Fax: +514-251-2617. E-mail: faardema @crfs.rtss.qc.ca
Contemporary cognitive models of obsessivecompulsive disorder (OCD) tend to focus on specific beliefs and appraisals in the development and maintenance of this disorder. The emphasis on specific beliefs and appraisals operating in OCD is derived directly from Beck’s cognitive specificity hypothesis, which holds that different psychological disorders are characterized by different dysfunctional beliefs (Beck, 1976). The application of Beck’s model of psychopathology has found its most systematic application to OCD in the work of Rachman (1997) and Salkovkis (1985, 1989), who hold that it is not the unwanted intrusive cognition that leads to distress and compulsive behaviours, but how the person appraises these thoughts in terms of personal significance or responsibility. Since then, the main focus of these appraisal models of OCD has # 2006 Taylor & Francis ISSN 1650-6073 DOI 10.1080/16506070600621922
been to identify other specific obsessivecompulsive beliefs proposed to be relevant to OCD (see Taylor, 2002). The application of Beck’s model of psychopathology to OCD is based on several studies that have shown obsessions to be similar in content to intrusive cognitions occurring in the normal population. The similarity in content between intrusions and obsessions would suggest that intrusive cognitions are not the essential element in the development of OCD, but rather, it is how the person appraises these thoughts as dictated by specific dysfunctional beliefs (Rachman & DeSilva, 1978; Salkovskis & Harrison, 1984). However, this argument regarding the normal nature of obsessions in terms of content may have been taken too far (Clark & O’Connor, 2004). Rachman and Hodgson (1980) already
VOL 35, NO 3, 2006
noted that a conceptualization of obsessions as intrusive cognitions was unsatisfactory, and while there may be similarities in content between intrusive cognitions and obsessions, there may be differences in the particular way that they arise. In particular, there may be important differences in the form and context of obsessions, which highlights the role of important processes operating in OCD that go beyond cognitive content considerations (O’Connor, 2002; Julien & O’Connor, 2005). An inference-based approach (IBA) to OCD (Aardema & O’Connor, 2003; O’Connor & Robillard, 1995, 1999) identifies a characteristic reasoning process associated with the occurrence of obsessions and, as such, is more concerned with the form and context of the obsession rather than its (ab)normal content. This is not a "general deficit model" where obsessions are solely the result of any particular cognitive dysfunction; rather, the IBA holds that the obsessional doubt, finds its justification in a wide variety of idiosyncratic narratives that contain inductive reasoning processes peculiar to OCD (O’Connor & Robillard, 1999). As such, the "intrusions" in OCD ("maybe my hands are contaminated"; "maybe I drove over someone with my car") inherit their persistence and reality value from reasoning processes associated with their occurrence rather than being the result of a specific belief. A central reasoning process that has been identified by the IBA model as relevant to the occurrence of obsessions is inverse inference. Inverse inference refers to an inverse type of reasoning where the person does not start out with the senses in reaching a conclusion ("I see there is mud on the floor, so people may have walked over it with dirty shoes"), but instead comes to infer a remote possibility without any actual indication of it being present or even in contradiction to what is seen or sensed ("People may have walked with muddy shoes on the floor, therefore the floor is dirty"). Another reasoning process closely related to inverse inference is a distrust of the senses. Often, obsessional doubts justify disregarding the senses in favour of going deeper into reality (e.g. "Even though my senses tell me nothing is there, I know by my intelligence that there is"), which in fact negates or ignores reality in favour of a hypothetical (and completely subjective) possibility. These
Inferential confusion and obsessive beliefs in OCD
139
reasoning processes, which lead a person with OCD to treat the obsessional doubt as a real probability, have broadly been termed "inferential confusion" (O’Connor, Aardema & Pe´lissier, 2005). Empirical investigations into OCD have so far found encouraging evidence for a role of "inferential confusion" operating in OCD. Psychometric studies have consistently shown substantial relationships between inferential confusion and obsessive-compulsive symptoms in non-clinical and clinical samples (Aardema, Kleijer, Trihey, O’Connor, & Emmelkamp, 2006; Aardema, O’Connor, Emmelkamp, Marchand, & Todorov, 2005; Emmelkamp & Aardema, 1999). Therapy specifically targeting inference processes in OCD leads to equivalent or greater improvement than "conventional" cognitive behaviour therapy (CBT) and exposure in vivo and response prevention (O’Connor et al., 2005). In addition, successful treatment outcome for OCD was found to be associated with changes in inferential confusion even if inference processes were not directly targeted in an OCD sample receiving CBT (Aardema, Emmelkamp, & O’Connor, 2005). Finally, experimental studies have highlighted the importance of reasoning processes operating in OCD, which leads people with OCD to replace certainty with doubt (Pe´lissier & O’Connor, 2002). The IBA model is not incompatible with an appraisal-based approach to OCD, whose primary concern is with the interpretation of the intrusion guided by specific obsessive beliefs. However, an inference and appraisal approach seem to target different stages in the obsessional sequence (Clark & O’Connor, 2004). Inference processes are primarily associated with the occurrence of obsessions whereas OCD relevant appraisals by definition occur in the aftermath of obsessions. In fact, the IBA model conceptualizes the initial "intrusion" in OCD as an inference about a possible, but improbable state of affairs, which comes about as the result of inferential confusion – a completely subjective inferential process that negates current reality (O’Connor, 2002). This intrusion may subsequently be appraised, but the appraisal model considers the occurrence of obsessional intrusions as a "given" phenomenon that requires little further explanation. However, if indeed
140
Aardema, O’Connor and Emmelkamp
inferential processes are relevant to the form and context of intrusions, it is quite possible that the relationship between obsessional beliefs and obsessive-compulsive symptoms can be accounted for by inference processes. For example, an obsession such as "I might have driven over someone with my car" may logically result in a need for certainty, elevated responsibility, attempts at control, and to give the thought importance; if this thought is experienced as realistic due to a confusion between reality and possibility (i.e. inferential confusion). In other words, if I really was convinced that, because I left the light on, the apartment block would burn down and everybody’s belongings and lodgings destroyed, it would seem realistic to feel anxious, guilty and responsible. As such, it has been suggested that some obsessive-compulsive beliefs and appraisals may follow logically and naturally from the intensity and reality value of the primary obsessional intrusion (Aardema & O’Connor, 2003). Deductive abilities seem intact in OCD (Pe´lissier & O’Connor, 2002; Reed, 1977), and hence valid deductive conclusions may be drawn from a faulty inductively generated premise. Due to the theoretical foundations of the appraisal model (i.e. intrusions as normal phenomena) overestimation of threat is generally also considered solely in the aftermath of intrusions. For example, the person has the intrusion "I may harm my child" and subsequently interprets this as "I am in danger of doing it" (Sookman, Pinard, & Beck, 2001, p. 112). However, considering more enduring anomalous processes as responsible for the initial intrusion would be theoretically incoherent for the appraisal model of OCD (which considers such intrusions as normal), whilst the IBA model has less issue with the notion that a cognitive bias towards overestimation of threat may play some role in the initial occurrence of intrusions and in deciding their content. In particular, for the IBA model the personalised content of the initial threatening intrusion (e.g. "I may have left the front door open") could be primed by a self-referent theme preceding its onset (e.g. "I’m just the sort of person who could leave front doors open"). Yet, the IBA model would be particularly interested in the reasoning processes by which this (mis)perception of threat comes about. In particular, it would argue that the
COGNITIVE BEHAVIOUR THERAPY
perception of threat in OCD is characterized by inferential confusion where the person generates the obsessional inference (or threat) in the absence of sense data to support it. Hence, it would argue that even if obsessions are often characterized by overestimation of threat (i.e. "I’m in danger") it is the element of inferential confusion ("I am in danger even though I see and sense nothing to support it") that constitutes the unique obsessional element. Regardless of the exact causal mechanisms involved in the obsessional sequence, the question arises as to what extent obsessive beliefs are related to OCD symptoms independently of inferential confusion. Several studies have found support for a relationship between inferential confusion and obsessivecompulsive symptoms while controlling for competing cognitive domains as well as negative mood states (Aardema, Kleijer et al., 2006; Aardema, O’Connor et al., 2005; Emmelkamp & Aardema, 1999). However, whether or not the relationship of obsessive beliefs with OCD symptomatology is independent of levels of inferential confusion remains as yet un-investigated. The goal of the present study was to establish whether inferential confusion could account for most of the relationships between obsessional beliefs and obsessive-compulsive symptoms. In line with our theoretical formulation, we expected that most of the relationships between belief domains and obsessive-compulsive symptoms could be accounted for by inferential confusion. In addition, a priori, we addressed a competing hypothesis that it is not inferential confusion, which accounts for the relationship between obsessive-compulsive beliefs and symptoms, but level of anxiety and overestimation of threat.
Method Recruitment and participants Participants in the study were recruited under the auspices of the OCD research program already in place at Centre de Recherche Fernand-Seguin. (For full details of sample characteristics see Aardema, O’Connor et al., 2005.) This recruitment included face-to-face diagnostic interview, and administration of a semi-structured interview (Brown, Di Nardo,
VOL 35, NO 3, 2006
& Barlow, 1994; Goodman et al., 1989). The total sample comprised 85 participants diagnosed with OCD (54 female, 31 male). The average age for the entire group was 37.6 years (SD511.9; range 17–59 years). Education levels were: 23.8% secondary education, 31.7% college education and 40% university education. The marital status was: 43.5% single, 28.6% married or cohabiting, and 12.7% separated or divorced.
Measures All participants were administered and completed the questionnaires described below. Inferential Confusion Questionnaire (ICQ; Aardema, O’Connor et al., 2005). This questionnaire measures several key aspects of inferential confusion, as formulated by O’Connor and Robillard (1995). Items reflect a tendency to distrust the senses and inverse inference, where the person infers a state of affairs in reality without any actual indication of it being present or even in contradiction to what is seen or sensed. The items of the ICQ15 are scored on a 5-point scale: 15strongly disagree, 25disagree, 35neutral, 45agree, 55strongly agree. Previous research has found the ICQ to be reliable (a50.85; Aardema, O’Connor et al., 2005). Padua Inventory Washington University Revision (PI-R; Burns, Keortge, Formea, & Sternberger, 1996). This is a comprehensive 39-item self-report inventory of obsessions and compulsions, based on the original version of the Padua Inventory (Sanavio, 1988). Items are rated on a 5-point scale
Inferential confusion and obsessive beliefs in OCD
(05not at all typical to 55very typical). The PI-R content dimensions relevant to OCD: (1) Obsessional thoughts (7 items), (2) Contamination (10 items), (3) Checking (10 items), (4) Dressing/grooming (3 items) and (5) Obsessional impulses (9 items). The total scale (a50.95) and the subscales are reliable (a50.75–0.91; Burns et al., 1996). Obsessive Belief Questionnaire (OBQ-44; Obsessive Compulsive Cognitions Working Group (OCCWG), 2005). This instrument has been developed collaboratively by the Obsessive Compulsive Working Group. The OBQ-44 is a shortened version of the OBQ-87 (OCCWG, 2003) whose scales have been derived through factor-analyses as opposed to the rationalistic generated scales of the OBQ87. It consists of 3 scales, namely (1) Responsibility/Overestimation of Threat, (2) Tolerance for Uncertainty/Perfectionism, and (3) Importance of Thoughts/Control of Thoughts. The subscales are reliable (a50.89– 0.93; OCCWG, 2005). For the current study, the OBQ-87 version was administered, and scored as OBQ-44, unless indicated otherwise. Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988). This is a 21item anxiety symptom checklist rating symptom intensity for the last week on a 0–3 scale (a50.91, Beck et al., 1988).
Results Descriptive statistics showing means and standard deviations of the questionnaires for the entire sample are shown in Table 1.
Table 1. Means and standard deviations of the ICQ, OBQ-44 and symptom measures (n585).
ICQ OBQ44-Total Responsibility/Overestimation of Threat Tolerance for Uncertainty/Perfectionism Importance of Thoughts/Control of Thoughts Padua Revised-Total Score Thoughts about harm Impulses about harm Contamination Checking Dressing/grooming Beck Anxiety Inventory
141
M
SD
49.1 188.4 66.8 78.2 43.4 63.2 10.7 3.5 18.0 21.4 5.7 20.3
12.0 53.8 25.0 20.8 17.1 24.0 6.1 4.8 11.1 9.5 4.0 13.2
142
COGNITIVE BEHAVIOUR THERAPY
Aardema, O’Connor and Emmelkamp
Table 2. Intercorrelations of the ICQ and OBQ belief domains (n585). OBQ44-T RT PC ICT ICQ
0.90*** 0.84*** 0.82*** 0.57***
RT
PC
– 0.60*** 0.64*** 0.71***
– – 0.55*** 0.29**
ICT – – – 0.41***
pv0.05, ** pv0.01, *** pv0.001. OBQ44-T5Obsessional Beliefs Questionaire-44 Total Score; RT5Responsibility/Threat; PC5Perfectionism/Intolerance of uncertainty; ICT5Importance given to thoughts/Control of thoughts; ICQ5Inferential Confusion Questionnaire.
Intercorrelations between the ICQ and OBQ-44 subscales We calculated interrelations between the ICQ and OBQ-subscales to establish the level of overlap between the different cognitive domains. Correlations are shown in Table 2. As expected, the relationships between the various cognitive domains are quite strong. The interrelationships between the OBQ belief subscales range from 0.55 to 0.64, and not surprisingly, are highly correlated with the OBQ total score. The ICQ shows discriminant validity with the OBQ subscales Perfectionism/Certainty and Importance/ Control of Thoughts, but is strongly related to the OBQ subscale Responsibility/Threat estimation. The magnitude of these intercorrelations indicates that zero order correlations alone would not be conclusive in determining which cognitive domain best accounts for the variance in obsessive-compulsive symptoms.
Zero-order and partial correlations of the ICQ and OBQ-44 with obsessivecompulsive symptoms The main purpose of the present study was to establish whether inferential confusion could account for the relationship between OBQ-44 belief domains and obsessive-compulsive symptoms. A powerful method to establish whether the relationship between variables is accounted for by a third variable is the calculation of partial correlations, which represents the relationship between 2 variables with the effects of a third variable removed (or held constant) (see Nunnally & Bernstein, 1994). If the partial correlation between 2 variables decreases substantially relative to the zero-order correlation then this relationship is said to be accounted for (although not necessarily explained) by a third variable.
We calculated zero-order and partial correlations (between brackets) between the OBQ belief domains and obsessive-compulsive symptoms, while controlling for inferential confusion (see Table 3). Additionally, for comparison purposes we calculated the zero-order and partial correlations (between brackets) between inferential confusion and obsessive-compulsive symptoms while controlling for obsessive beliefs. The results confirmed our expectations that inferential confusion can largely account for the relationship between OCD beliefs and obsessive-compulsive symptoms. While zeroorder correlations of the OBQ-44 beliefs with obsessive-compulsive symptoms are substantial, these relationships decreased considerably when controlling for inferential confusion. However, a few significant relationships remain when controlling for inferential confusion. The subscale Perfectionism/ Certainty remained significantly related to the total score of the PI-R total and the subscale checking compulsions. As well, the subscale Importance/Control of thoughts remained significantly related to obsessions about harm. In contrast to these results, the relationship between inferential confusion and obsessivecompulsive symptoms when simultaneously controlling for all of the 3 OBQ belief domains showed a far less dramatic decrease. Most of the relationships between inferential confusion and obsessive-compulsive symptoms remained substantial and significant when controlling for obsessive beliefs.
Controlling for overestimation of threat and anxiety Inferential confusion as measured by the ICQ has an overlap with the construct of overestimation of threat (Aardema, O’Connor, Emmelkamp, Marchand, Todorov, 2005).
VOL 35, NO 3, 2006
Inferential confusion and obsessive beliefs in OCD
143
Table 3. Zero-order correlations and partial correlations (in parentheses) of inferential confusion and OBQ belief domains with obsessive-compulsive symptoms (n585). (Controlled for ICQ) OBQ44-T PI-R Total Zero-order correlations Controlled PI-R Obsessions Zero-order correlations Controlled PI-R-Impulses Zero-order correlations Controlled PI-R Checking Zero-order correlations Controlled PI-R Contamination Zero-order correlations Controlled PI-R Dressing Zero-order correlations Controlled
RT
PC
ICT
ICQ (Controlled for OBQ domains)
0.48*** (0.18)
0.47*** (0.06)
0.39** (0.27)*
0.34* (0.09)
0.52*** (0.43)***
0.57*** (0.24)*
0.61*** (0.20)
0.32** (0.10)
0.50*** (0.33)*
0.72*** (0.46)***
0.18 (0.10)
0.29* (0.20)
0.20 (0.11)
0.10 (20.16)
0.26* (0.06)
0.25* (0.11) 0.32** (0.03)
0.19 (20.02) 0.34** (0.06)
0.34** (0.26)*
0.28* (0.01)
0.28* (20.07)
0.22* (0.08)
0.19 (20.09)
0.11 (20.20)
0.29** (0.19)
0.21 (0.01) 0.08 (20.03)
0.38*** (0.36)*** 0.12 (0.14)
pv0.05, ** pv0.01, *** pv0.001. OBQ44-T5Obsessional Beliefs Questionaire-44 Total Score; RT5Responsibility/Threat; PC5Perfectionism/Intolerance of uncertainty; ICT5Importance given to thoughts/Control of thoughts; ICQ5Inferential Confusion Questionnaire; PI-R5Padua Inventory Revised.
Thus, it could be argued that some of the results in the present study were attenuated by the overlap between the ICQ and overestimation of threat. In particular, the overlap between overestimation of threat and inferential confusion may be responsible for the ability of inferential confusion to account for the relationship between obsessive beliefs and obsessive-compulsive symptoms. We examined this possibility by first investigating whether the item set of the ICQ could be empirically distinguished from overestimation of threat through factor analyses. We performed a principal component analyses on the items of the ICQ and the original OBQ-87 subscale overestimation of threat, followed by varimax rotation in order to extract 2 independent factors. The rationale for using the original OBQ-87 overestimation of threat scale was to ensure an adequate representation of items reflecting overestimation of threat (in line with the hypothesis) rather than the mixed scale of the OBQ-44 consisting of both responsibility and overestimation of threat items. Results indicated 1 large first factor with an eigenvalue of 11.0 explaining 38.0% of the variance, followed by a second factor with an eigenvalue of 2.3
explaining an additional 7.9% of variance. Subsequent factors had eigenvalues of respectively 1.8, 1.5, 1.2, 1.1 and 1.0. Thus, examination of the scree plot indicated support for extraction of 2 factors. Factor loadings on a 2-factor solution after varimax rotation are shown in Table 4. Salient loadings (>0.40) are shown in bold. As can be seen in Table 4, the first factor mostly contained highest loadings from the items of the ICQ, whereas the items of the overestimation of threat scale had most of their highest loadings on the second factor. Only 1 item of the ICQ had its primary loading on the construct of overestimation of threat (ICQ item 5), and only 3 items from the overestimation of threat scale had their primary loading on the construct of inferential confusion (OBQ item 6, 39, 52). An additional 2 items of the overestimation of threat scale had secondary loadings (w0.40) on the construct of inferential confusion (OBQ items 68, 79). None of the items of the ICQ had secondary loadings on the construct of overestimation of threat. In order to determine whether the varimax rotation was truly reflective of the underlying factor structure we also performed an oblique
144
Aardema, O’Connor and Emmelkamp
COGNITIVE BEHAVIOUR THERAPY
Table 4. Factor loadings after varimax rotation with abbreviated item content. Factor 1: Inferential Confusion ICQ9 ICQ8 ICQ12 ICQ11 ICQ10 ICQ1 ICQ15 ICQ2 ICQ7 ICQ13 ICQ6 ICQ14 OBQ39 ICQ3 OBQ6 ICQ4 OBQ52 OBQ82 OBQ40 OBQ50 OBQ80 OBQ61 OBQ79 OBQ16 OBQ30 OBQ72 ICQ5 OBQ68 OBQ9
There are many invisible dangers Even without proof my imagination convinces me otherwise My imagination makes me lose confidence in what I perceive Knowing a problem exists without visible proof Just a thought is enough proof of danger More convinced about what might be than what is seen Reacting to something that might be as if it is happening Inventing dangers that might be without seeing Idea of danger without obvious reason In spite of evidence, feeling that danger will occur Something not safe, because things are not as they appear More afraid of unseen than seen Avoiding serious problems requires constant effort Knowing there’s danger without feeling the need to look I think things around me are unsafe One can never know something is safe on appearances I believe the world is a dangerous place When things go wrong it’s likely to have terrible effects Small things turn into big problems in my life Not taking precautions increases the risk of an accident When things go well, something bad will follow I’m more likely than others to cause harm Ordinary experiences in my life are full of risk Minor annoyances seem like disasters to me Bad things are more likely to happen to me than others Harmful events will happen unless I’m careful Thinking there is danger and taking precautions Even when I’m careful, I often think bad things will happen I am more likely to be punished than others
0.73 0.71 0.66 0.65 0.65 0.64 0.64 0.58 0.58 0.57 0.56 0.56 0.56 0.53 0.52 0.43 0.41 0.09 0.02 0.33 0.09 0.03 0.42 0.16 0.28 0.35 0.28 0.46 0.33
Factor 2: Threat Estimation 0.01 0.25 0.30 0.17 0.38 0.21 0.33 0.29 0.29 0.30 0.24 0.02 0.45 0.08 0.45 0.17 0.22 0.81 0.69 0.69 0.81 0.67 0.65 0.64 0.63 0.58 0.57 0.53 0.48
Salient loadings (w0.40) are indicated in bold. OBQ5Obsessional Beliefs Questionaire; ICQ5Inferential Confusion Questionnaire.
rotation. Oblique rotation resulted in very similar findings. with only 2 items of the inferential confusion questionnaire having their highest loading on a second factor representing the construct of overestimation of threat, and none of the items of the overestimation of threat subscale having their highest loading on the first factor representing the construct of inferential confusion. Thus, these results established that the construct of inferential confusion could be clearly separated from overestimation of threat through factor analyses. Moreover, since both factors came about through varimax rotation (and so were uncorrelated), the independent correlation of each factor with OCD symptoms should reveal whether the ICQ still accounted for the relationship between OBQ belief
domains without any overlap with overestimation of threat. We calculated the correlations between each of the independent factors derived through varimax rotation (using factor-scores) and obsessive-compulsive symptoms. Both overestimation of threat and inferential confusion were independently related to obsessive-compulsive symptoms. The construct of inferential confusion was independently significantly related to the PI-R total score (r50.41; pv0.001), obsessions about harm (r50.60; pv0.001) and washing compulsions (r50.39; pv0.001). No significant relationships were found with the other subscales of the PI-R. The construct of overestimation of threat was independently significantly related to the PI-R total score (r50.44; pv0.001),
VOL 35, NO 3, 2006
Inferential confusion and obsessive beliefs in OCD
obsessions about harm (r50.47; pv0.001), obsessional impulses (r50.24; p50.03) and checking compulsions (r50.39; pv0.001). No significant relationships were found with the other subscales of the PI-R. Since overestimation of threat appears to be independently related to obsessive-compulsive symptoms these results do not exclude the possibility that the ability of inferential confusion to account for the relationships between OBQ belief domains and obsessivecompulsive symptoms may in part have been caused by its overlap with overestimation of threat. However, it can be questioned whether this second factor actually represents overestimation of threat. Overestimation of threat is often considered to be a general vulnerability factor in anxiety disorders (Beck, Emery & Greenberg, 1985), and as such, this factor may be more representative of anxious mood rather than form a particular cognitive bias. In order to test this hypothesis we calculated the relationship between overestimation of threat (using factor-scores) and obsessive-compulsive symptoms while controlling for anxious mood (BAI). Results of these partial correlational analyses showed that the construct of overestimation of threat only remained significantly related to checking compulsions (r50.27; p50.02), while no significant relationships remained with any of the other obsessive-compulsive symptoms. In contrast, the construct of inferential confusion remained significantly related with obsessivecompulsive symptoms overall (r50.26; p50.03), obsessions about harm (r50.48; pv0.001) and washing compulsions (r50.26; p50.03) when controlling for anxious mood. Therefore, the competing hypothesis that overestimation of threat may be responsible for the ability of inferential confusion to account for the relationship between beliefs and obsessive-compulsive symptoms is not supported.
Discussion The main goal of the present study was to establish whether inferential confusion could account for most of the relationships between belief domains and obsessive-compulsive symptoms. Results indicated that inferential confusion accounted substantially for the relationships between obsessive-compulsive
145
beliefs and obsessive-compulsive symptoms. In contrast, the relationship between inferential confusion and obsessive-compulsive symptoms remained substantial when controlling for obsessive beliefs. Hence, these results provide support for the view that the relationship between obsessive beliefs and obsessivecompulsive symptoms can be accounted for by inferential confusion. A competing hypothesis, which attributed the current results to the overlap between inferential confusion and overestimation of threat, was not supported. After separating the shared variance of both constructs with obsessive-compulsive symptoms, and controlling for anxious mood, inferential confusion was significantly related to several obsessivecompulsive symptoms, while most of the relationships between overestimation of threat and obsessive-compulsive symptoms disappeared. In other words, overestimation of threat was found not to be a viable competing hypothesis for the current results. Rather, it appeared that the independent relationship between overestimation of threat and obsessive-compulsive symptoms was due to an overlap of overestimation of threat with anxious mood. It should be noted that these results do not provide evidence for any causal mechanisms proposed to be operating in OCD according to an inference-based approach. However, the results are consistent with our contention that obsessive-compulsive beliefs logically follow from the inferential confusion process associated with the occurrence of "intrusions" or obsessions. Cognitive elaboration via appraisal of intrusions or obsessions may certainly reinforce obsessive-compulsive symptoms. Rachman (2003) has recently drawn attention to the role of appraisals in generating fearful contexts for the obsession to occur, and resulting in its increased persistence and frequency. However, the present results seem to indicate that inferential confusion may be a more critical factor in accounting for most OCD symptoms than obsessive beliefs and appraisals. It should be recognized that multidimensional studies investigating the relationship between cognitive measures with obsessivecompulsive symptoms remain a challenge, since partial correlations cannot completely eliminate all competing hypotheses. For
146
Aardema, O’Connor and Emmelkamp
example, the current study did not control for depressive mood, which has been found to account for a portion of the variance in some obsessive-compulsive symptoms (Emmelkamp & Aardema, 1999). While inevitably alternative explanations for the current results remain, it should be noted that the current study addresses a variety of competing hypothesis in an attempt to isolate cognitive domains that are often confounded with each other. In this regard, a recent study investigating an even larger number of statistically independent cognitive domains and their relationship OCD symptoms as well as controlling for both anxiety and depression has continued to highlight the importance of inference processes operating in OCD (Aardema, Radomsky, O’Connor & Julien, in press). However, further research is needed to determine the relative contribution of inferential confusion and diverse negative mood states, such as depression, in explaining the relationships between obsessive-compulsive beliefs and obsessive-compulsive symptoms. As well, there may be other cognitive variables operating in OCD outside of both belief domains and inferential confusion, and current results remain tentative due to the limitations of correlational data. While the importance of psychometric investigations should not be underestimated (Taylor, 2002) causal pathways in the development and maintenance of OCD will ultimately need to be addressed through more experimental designs directly manipulating variables. In particular, the inferential confusion model would predict that people with OCD would be more affected than (non)clinical controls by subjective possibilitybased information that negates reality-based information. This would account for the persistence and occurrence of obsessional inferences in the absence of (or despite) proof to the contrary. In particular, studies that link such experimental and psychometric data could more concretely address current limitations.
References Aardema, F., Emmelkamp, P. M. G., & O’Connor, K. (2005). Inferential confusion, cognitive change and treatment outcome in obsessivecompulsive disorder. Clinical Psychology and Psychotherapy, 12, 337–345.
COGNITIVE BEHAVIOUR THERAPY
Aardema, F., Kleijer, T. M. R., Trihey, M., O’Connor, K., & Emmelkamp, P. (2006). Inference processes, schizotypal thinking and obsessive-compulsive disorder. (submitted). Aardema, F., & O’Connor, K. (2003). Seeing white bears that are not there: inference processes in obsessions. Journal of Cognitive Psychotherapy, 17, 23–37. Aardema, F., O’Connor, K., Emmelkamp, P., Marchand, A., & Todorov, C. (2005). Inferential confusion and obsessive-compulsive disorder: the Inferential Confusion Questionnaire. Behaviour Research and Therapy, 43, 293–308. Aardema, F., Radomsky, A. S., O’Connor, K. P., & Julien, D. (2006). Inferential confusion, obsessive beliefs and obsessive-compulsive symptoms: a multidimensional investigation of independent cognitive domains. (submitted). Beck, A. T. (1976). Cognitive Therapy and the Emotional Disorders. New York: International Universities Press. Beck, A. T., Emery, G., & Greenberg, R. L. (1985). Anxiety Disorders and Phobias: A Cognitive Perspective. New York: Basic Books. Beck, A. T., Epstein, N., Brown, G., & Steer, R. A. (1988). An inventory for measuring clinical anxiety: psychometric properties. Journal of Consulting and Clinical Psychology, 56, 893–897. Brown, T. A., DiNardo, P. A., & Barlow, D. H. (1994). Anxiety disorders schedule for DSMIV. Boulder, CO: Graywind Publications. Burns, G. L., Keortge, S. G., Formea, G. M., & Sternberger, L. G. (1996). Revision of the Padua Inventory of obsessive compulsive disorder symptoms: distinctions between worry, obsessions, and compulsions. Behaviour Research and Therapy, 14, 163–173. Clark, D. A., & O’Connor, K. (2004). Thinking is believing: ego-dystonic intrusive thoughts in obsessive-compulsive disorder. In D. A. Clark (Ed.), Unwanted Intrusive Thoughts in Clinical Disorders. New York: Guilford Press. Emmelkamp, P. M. G., & Aardema, F. (1999). Metacognition, specific obsessive-compulsive beliefs and obsessive-compulsive behavior. Clinical Psychology and Psychotherapy, 6, 139–145. Goodman, W. K., Price, L. H., Rasmussen, S. A., Mazure, C., Delgado, P., Heninger, G. R., & Charney, D. S. (1989). The Yale-Brown Obsessive Compulsive Scale. II. Validity Archives of General Psychiatry, 46, 1012–1016. Julien, D., & O’Connor, K. P. (2005). Intrusive thoughts in non-clinical populations: an overview of the literature. (submitted). Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric Theory. New York: McGrawHill Inc. Obsessive Compulsive Cognitions Working Group (2003). Psychometric validation of the Obsessive Beliefs Questionnaire and the Interpretation of Intrusions Inventory: Part I. Behaviour Research and Therapy, 41, 863–878.
VOL 35, NO 3, 2006
Obsessive Compulsive Cognitions Working Group (2005). Psychometric validation of the Obsessive Belief Questionnaire and Interpretation of Intrusions Inventory – Part 2: factor analyses and testing of a brief version. Behaviour Research and Therapy, 43, 1527–1542. O’Connor, K. P. (2002). Intrusions and inferences in obsessive-compulsive disorder. Clinical Psychology and Psychotherapy, 9, 38–46. O’Connor, K., Aardema, F., Bouthillier, D., Fournier, S., Guay, S., Robillard, S., Pe´lissier, M. C., Landry, P., Todorov, C., Tremblay, M., & Pitre, D. (2005). Evaluation of an inference based approach to treating obsessive-compulsive disorder. Cognitive Behaviour Therapy, 43, 1–16. O’Connor, K. P., & Robillard, S. (1995). Inference processes in obsessive-compulsive disorder: Some clinical observations. Behaviour Research and Therapy, 33, 887–896. O’Connor, K. P., & Robillard, S. (1999). A cognitive approach to modifying primary inferences in obsessive-compulsive disorder. Journal of Cognitive Psychotherapy, 13, 1–17. O’Connor, K., Aardema, F., & Pe´lissier, M. C. (2005). Beyond Reasonable Doubt: Reasoning Processes in Obsessive-Compulsive and Related Disorders. Chichester, UK: Wiley.
Inferential confusion and obsessive beliefs in OCD
147
Pe´lissier, M. C., & O’Connor, K. P. (2002). Deductive and inductive reasoning in obsessive-compulsive disorder. British Journal of Clinical Psychology, 41, 5–27. Rachman, S. (2003). The Treatment of Obsessions. New York: Oxford University Press. Rachman, S. J., & DeSilva, P. (1978). Abnormal and normal obsessions. Behaviour Research and Therapy, 16, 233–248. Rachman, S., & Hodgson, R. (1980). Obsessions and Compulsions. Englewood Cliffs, NJ: Prentice Hall. Reed, G. F. (1977). Obsessional cognition: performance on two numerical tasks. British Journal of Psychiatry, 130, 177–183. Salkovskis, P. M., & Harisson, J. (1984). Abnormal and normal obsessions: a replication. Behaviour Research and Therapy, 22, 549–552. Sanavio, E. (1988). Obsessions and compulsions: the Padua Inventory. Behaviour Research and Therapy, 26, 169–177. Sookman, D., Pinard, G., & Beck, A. T. (2001). Vulnerability schemas in obsessive compulsive disorder. Journal of Cognitive Psychotherapy, 15, (2), 109–130. Taylor, S. (2002). Cognition in obsessive-compulsive behaviour. An overview. In R. O. Frost, G. Steketee (Eds), Cognitive Approaches to Obsessions and Compulsions: Theory, Assessment and Treatment. Oxford: Elsevier.