Panic patients reveal idiographic associations between anxiety symptoms and catastrophes in a semantic priming task

https://doi.org/10.1016/j.brat.2006.02.007Get rights and content

Abstract

In the present study, a modified semantic priming paradigm was used to test whether panic patients more strongly associate catastrophes with anxiety symptoms than nonclinical subjects. Subjects named catastrophic target words (e.g. infarction) and target words neutral to anxiety themes (e.g. weekend) that followed auditive prime sentences immediately (i.e.0 ms) or with a delay (i.e. 1500 ms). Prime sentences described the perception of anxiety symptoms (e.g. You feel tense) or sensations neutral to anxiety (You feel relaxed). Consistent with an earlier study [Schniering C.A., & Rapee, R.M. (1997). A test of the cognitive model of panic: Primed lexical decision in panic disorder. Journal of Anxiety Disorders, 11, 557–571] the two groups did not differ if semantic priming effects were calculated in the usual way, i.e. by averaging across identical stimuli for all subjects. As expected, however, panic patients demonstrated stronger semantic priming effects for catastrophes immediately following prime sentences if priming effects were calculated for idiographically selected stimuli. The latter result indicated stronger automatic associations between idiographic anxiety symptoms and catastrophes in panic patients consistent with the cognitive model of panic disorder (Clark, D.M. (1986). A cognitive approach to panic. Behaviour Research and Therapy, 24, 461–470). The restriction of stronger associations in panic patients to idiographic stimuli is explained from an evolutionary perspective.

Introduction

According to cognitive theories of panic disorder (Beck, Emery, & Greenberg, 1985; Clark, 1986) panic patients associate harmless somatic sensations with acute, catastrophic outcomes. Typically, a patient spontaneously misinterprets an autonomic reaction, especially an anxiety symptom like palpitation as, for example, an impending heart attack, or nausea as a symptom of fainting. Indeed, a number of studies (Clark et al. (1988), Clark et al. (1997); Harvey, Richards, Dziadosz, & Swindell, 1993; McNally & Foa, 1987; Richards, Austin, & Alvarenga, 2001) have consistently demonstrated a stronger tendency of panic patients to supply catastrophic interpretations (e.g. heart attack) in response to descriptions of bodily sensations in interpretation questionnaires (e.g. You notice your heart beating quickly and pounding. Why?).

However, catastrophic biases as assessed by self-report measures might merely reflect a panic patient's attempt to explain panic symptomatology without being causal for the panic attacks. Moreover, self-report measures are subject to distortions by memory biases. A more fundamental critique is that the assessment of the catastrophic bias with questionnaires is artificial with respect to naturally occurring catastrophic cognitions. That is, the assumed catastrophic misinterpretations during a panic attack will not result from deliberate thinking, but instead are likely to come to mind very quickly or reflexively (Clark et al. (1988), Clark et al. (1997)). In other words, panic patients often seem to associate catastrophes with somatic symptoms automatically.

A methodology that does not rely on the self-report of subjects and is especially suitable for investigating automatic associations is the semantic priming task (Meyer & Schvaneveldt; 1971; Meyer, Schvaneveldt, & Ruddy, 1975). In the typical semantic priming task subjects have to silently read a prime word on a computer screen and then name a second one, the target word, which replaces the prime word (see Neely, 1991, for a review). A reliable effect is that responses to targets are faster if prime and target are semantically related (e.g. bread – butter) than if they are unrelated (e.g. chair – butter). Importantly, Neely (1977) demonstrated that semantic priming can be observed even if the target follows the prime so quickly (i.e. less than 250 ms after the prime appeared) that subjects are unable to use the prime to generate conscious expectancies about the target, which strongly suggests automatic processing (Posner & Snyder, 1975). That is, since prime and target of the related condition are associated in semantic memory, activation is assumed to spread from the concept node of the prime to the concept node of the target with virtually no time lapse and without the subject's intention. As a consequence, the activation level of the target word will initially (i.e. at the time of target presentation) be higher if prime and target are related than if they are not related, resulting in faster responses in the former than in the latter case.

Clark et al. (1988) used a semantic priming paradigm to demonstrate that panic patients automatically associate catastrophes with anxiety symptoms. Indeed, panic patients named catastrophic words (e.g. “dying”) following descriptions of anxiety symptoms faster than neutral words (e.g. “excited”), whereas healthy subjects did not. However, in this investigation participants were explicitly encouraged to make inferences about the target by the way prime sentences were formulated (e.g. prime: “If I had palpitations I could be”; target: “dying”), suggesting that priming effects result from controlled processing. Moreover, presentation times of primes and targets were not reported, leaving the possibility open that the time allowed for prime processing was indeed long enough for subjects to consciously generate expectancies about the following target. These methodological limitations make it difficult to definitely interpret higher semantic priming in panic patients as a result of automatic in contrast to controlled processing.

To overcome these methodological limitations, Schniering and Rapee (1997) used only single words as primes which subjects merely had to read silently, and controlled time for prime processing, i.e. target presentation started 2000 or 240 ms later than prime presentation. Although they were able to demonstrate semantic priming effects for catastrophic words in panic patients and normal subjects even in the experimental condition with the short time interval for prime processing, the groups did not differ significantly. Thus, under conditions strongly suggesting that priming effects were due to automatic processing there was no evidence of a stronger association between anxiety symptoms and catastrophes in panic patients.

However, it is very likely that the assumed panic-specific associations between symptoms and catastrophes are idiosyncratic and limited in number (Rapee, 1993). That is, there is probably only a small number of associations that are strengthened by the panic disorder. Moreover, which symptom is associated with which catastrophe varies from patient to patient. This situation makes it difficult to demonstrate stronger associations by the commonly used method of calculating semantic priming effects, i.e. by averaging reaction times across all word pairs that were originally chosen by experimenters independently of their idiographic emotional significance. Therefore, in the present study panic patients and nonclinical subjects participated in a modified semantic priming task (Swinney, 1979) which allowed investigation of automatic (i.e. an inter-stimulus interval, ISI, between prime and target of 0 ms) and controlled processing (i.e. ISI=1500 ms). Afterwards both groups explicitly judged the semantic relatedness of the presented prime-target pairs as quickly as they could (speeded yes–no relatedness decision task). It was assumed that stimuli which were more quickly recognized as semantically related would be more strongly associated in semantic memory. A selection of pairs of anxiety symptoms and catastrophes that were judged most quickly as related will then more likely comprise panic-specific stimuli with a heightened strength of association for the patients. Thus, differences in semantic priming for catastrophic targets between panic patients and nonclinical subjects should be enhanced if these priming effects were individually calculated for the most strongly related stimuli.

The speeded yes–no relatedness decision task just described seemed to us to be more appropriate for the purpose of out-filtering associated stimuli than a rating task, because the latter permits more deliberate processing than the former. Nevertheless, to demonstrate that faster relatedness decisions indeed represent stronger relatedness of stimuli the relatedness of test-relevant prime target pairs was also rated by subjects.

The central hypothesis of the present investigation was that panic patients should demonstrate stronger automatic semantic priming effects (i.e. at ISI=0 ms) than nonclinical subjects for idiographically selected catastrophic targets, whereas no differences were expected for these words when semantic priming effects were calculated in the usual way by averaging across all (unselected) catastrophic targets. Moreover, no significant differences between groups were expected for idiographically selected neutral targets, whether for automatic or for controlled semantic priming (i.e. at ISI=1500 ms). No specific hypothesis was made about differences between the two groups in controlled semantic priming (i.e. at ISI=1500 ms) for catastrophic targets.

Section snippets

Subjects

Forty-eight panic patients and 32 nonclinical controls, all native speakers, participated in the study. The patients were in treatment as outpatients of the Department of Clinical Psychology of the Ruhr-University of Bochum, Germany. They were included in the study according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association, 1994) and tested before psychotherapeutic treatment. Diagnoses were based on the Structured Clinical

Subject characteristics

Mean scores of panic patients were significantly higher on the ACQ (M=2.23 (SD=.66) vs. M=1.53 (SD=.31); t(78)=5.56, p<.001), the BSQ (M=3.10 (SD=.73) vs. M=2.13 (SD=.59); t(78)=6.28, p<.001), the ASI (M=29.94 (SD=9.68) vs. M=17.72 (SD=9.75); t(78)=5.53, p<.001), the MI (for situations in which they were alone: M=2.67 (SD=.95) vs. M=1.35 (SD=.39); t(78)=7.44, p<001; for situations accompanied by a trusted person: M=1.98 (SD=.75) vs. M=1.57 (SD=.64); t(78)=2.52, p=.014), and on the questions

Discussion

Consistent with Schniering and Rapee (1997), in the present study panic patients and nonclinical subjects showed no differences in semantic priming for catastrophic targets whether at the automatic (i.e. at ISI=0 ms) or at the strategic level (i.e. at ISI=1500 ms) when priming effects were calculated in the usual way, i.e. by averaging across identical stimuli for all subjects (nomothetic semantic priming effects). The two groups also did not differ in semantic priming for catastrophic targets

Acknowledgments

We thank Norman Ehlert for carefully running experimental test sessions and Dagmar Meister for her help in recruiting the patients. We are also grateful to two anonymous reviewers for their helpful comments. The research for this manuscript was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG).

References (34)

  • Beringer, J. (1993b). Experimental Run Time System. Version 3.00. Benutzer...
  • D.L. Chambless et al.

    Assessment of fear of fear in agoraphobics: The Body Sensations Questionnaire and the Agoraphobic Cognitions Questionnaire

    Journal of Consulting and Clinical Psychology

    (1984)
  • D.M. Clark et al.

    Tests of a cognitive theory of panic

  • D.M. Clark et al.

    Misinterpretation of body sensations in panic disorder

    Journal of Consulting and Clinical Psychology

    (1997)
  • Ehlers, A. (1986). Angst-Sensitivitäts-Index (ASI). Unveröffentlichtes...
  • Ehlers, A., Margraf, J., & Chambless, D. (1993). Diagnostische Batterie für Paniksyndrome und Agoraphobie. Weinheim:...
  • First, M. B., Gibbon, M., Spitzer, R. I., & Williams, J. B. W. (1996). Users guide for the structured clinical...
  • Cited by (16)

    • Anxiety Sensitivity and Arousal Symptom Implicit Association Task Performance: An Event-Related Potential Study of Cognitive Processing of Anxiety-Relevant Stimuli

      2021, Journal of Affective Disorders
      Citation Excerpt :

      For example, AS is positively associated with attentional bias towards panic-related words (Stewart, Conrod, Gignac, & Pihl, 1998) and misinterpretation of ambiguous internal (McNally, Hornig, Hoffman, & Han, 1999; Richards, Austin, & Alvarenga, 2001) and external stimuli (Richards et al., 2001; Teachman, 2005). Further, individuals with elevated AS interpret benign bodily sensations (e.g., elevated heart rate) as threatening (Schneider & Schulte, 2007), and AS predicts distress following experimental induction of physiological symptoms of anxiety (Farris & Zvolensky, 2016; Schmidt & Zvolensky, 2007). However, processes that contribute to the association between AS and biased cognitive processing are largely unknown.

    • A systematic review of the word sentence association paradigm (WSAP)

      2019, Journal of Behavior Therapy and Experimental Psychiatry
    • Assessment of automatic associations with bodily sensations and agoraphobic situations in panic disorder

      2016, Journal of Behavior Therapy and Experimental Psychiatry
      Citation Excerpt :

      For example, the tested samples differed in panic-related severity, and panic-related associations might have been more accessible in those patients with more severe symptoms. Furthermore, the stimuli used differed in their ecological validity and results partly depended on analyses using idiographically selected stimuli (Schneider & Schulte, 2007). This inconsistency makes it difficult to evaluate the role of panic-related associations adequately.

    • An fMRI-study on semantic priming of panic-related information in depression without comorbid anxiety

      2014, Psychiatry Research - Neuroimaging
      Citation Excerpt :

      The same result was found only for negative words in an earlier study (Sass et al., 2014) supporting the assumption that negative information captures more attentional resources in patients so that there are fewer processing resources available for successful task execution, i.e., negative information and, additionally, panic-related information are not relevant for successful accomplishment of the task, but patients are not able to ignore this information, leading to interference with the requested response (Sass et al., 2014). In addition, earlier studies on panic disorder showed that PD patients associate anxiety symptoms and catastrophes, suggesting stronger associations between idiographic anxiety symptoms and catastrophes (Schneider and Schulte, 2007), i.e., there might be an information-processing bias towards threatening information in PD (Teachman et al., 2007). Hence, as the depressed patients also showed a bias towards threatening targets during semantic priming, it might be speculated that a similar mechanism occurs.

    • Priming associations between bodily sensations and catastrophic misinterpretations: Specific for panic disorder?

      2010, Behaviour Research and Therapy
      Citation Excerpt :

      Using a priming paradigm, these authors demonstrated that stronger associations between idiographically related anxiety symptoms and catastrophes predicted a smaller reduction in anxiety sensitivity in panic patients in response to a brief cognitive–behavioural treatment (Schneider & Schulte, 2008, p. 568). A final element for discussion concerns the observation that group differences in panic priming only emerged after idiographical selection of prime-target pairs in the study of Schneider and Schulte (2007) whereas reliable differences emerged in our studies for an unselected set of prime-targets. To maximize the chance of observing the predicted (differences in) priming effects in the present two studies, all participants from the clinical groups were included on a moment at which they had been actively seeking treatment but did not yet enter the treatment programme.

    View all citing articles on Scopus
    View full text