فرآیندهای حافظه ضمنی و آشکار در بیماران هراس در اقدامات رفتاری و الکتروفیزیولوژیک منعکس شده

Implicit and explicit memory processes for panic-relevant and neutral word stimuli were examined in 16 panic patients and 16 healthy participants matched for sex, age, and education, using behavioral and electrophysiological measures. In the study phase, panic-relevant and neutral words were presented and the level of processing was varied by requiring either shallow (orthographic) or deeper (syntactic) processing. Implicit memory was tested with a lexical decision task, explicit memory with a recognition task. Panic patients and healthy participants did not differ in behavioral (response time) or event-related brain potential (ERP) measures of implicit memory. However, panic patients deviated from healthy participants in the recognition test, an explicit memory test. Although recognition of panic words was overall worse compared to neutral words, panic patients compared to healthy participants exhibited enhanced discrimination scores and faster reaction times for panic words. The level of processing manipulation had comparable effects on patients and healthy participants. While neither behavioral nor electrophysiological measures provided evidence for an implicit memory bias in panic patients, behavioral measures confirmed an explicit memory bias in panic patients for panic-relevant stimuli.

Vicious circle (Clark (1986) and Clark (1988)) and cognitive network (Foa & Kozak, 1993) models of panic disorder (PD) suggest that an abnormal processing of certain bodily symptoms is crucial for the development or maintenance of the disorder (Pauli et al., 1997). In PD, studies using information-processing paradigms revealed attentional, memory, interpretive, and interoceptive biases mainly for the processing of threatening, body-related stimuli (see the review by McNally, 1994). In PD, memory biases were not found consistently, although the support for an explicit memory bias is stronger than for an implicit memory bias. Coles and Heimberg (2002) summarized that nine out of 15 studies (60%) found an explicit bias, while only two out of five studies (40%) revealed an implicit memory bias. Implicit memory measures were word stem completion tasks (Cloitre, Shear, Cancienne, & Zeitlin, 1994; Lundh, Czyzykow, & Öst, 1997; Rapee, 1994) or the white noise judgment paradigm (Amir, McNally, Riemann, & Clements, 1996). Coles and Heimberg (2002) concluded that further research specifically on implicit memory bias in PD is needed. Because discrepancies in findings may be related to the nature of the encoding task, they also suggested that it would be worthwhile to examine the effects of variations in encoding activities. Variations in encoding may be accomplished by the level of processing manipulations, which have a clear effect on explicit memory as assessed with a recognition task (e.g., Paller, Kutas, & McIsaac, 1995). Although never examined directly, Coles and Heimberg (2002) speculated that in anxiety patients shallow processing during the study phase is unlikely to produce an explicit memory bias, while conceptual or deep processing presumably facilitates explicit memory biases. The present study systematically varied depth of processing during the study phase in order to examine this prediction in PD. Word stem completion is the most frequently used measure of implicit memory biases in anxiety disorders. However, this measure has the disadvantage that participants have to produce the stimulus they fear (Coles & Heimberg, 2002). Anxiety patients normally avoid feared stimuli, and failures to obtain an implicit memory bias may be due to the patients’ hesitation to produce the stimulus. The white noise paradigm and lexical decision tasks circumvent this disadvantage because participants do not have to produce stimuli. While the former was successfully used in PD patients (Amir et al., 1996), the latter has not been studied in PD patients. In the present study, a lexical decision test was realized to examine an implicit memory bias in PD patients. Participants had to decide whether a presented letter string constitutes a word or a non-word. Implicit memory should be reflected in faster reaction times (RTs) for studied words compared to new words (e.g., Paller et al., 1995). Previous studies with healthy participants (Paller et al., 1995) demonstrated that the processing level at encoding did not affect later lexical decisions; both deep and shallow processing during encoding caused faster RTs in lexical decisions compared to new words that were not presented during the prior study phase. However, event-related brain potentials (ERPs) recorded during lexical decisions were affected by the level of processing manipulations. ERPs 500–900 ms after word onset were more positive for deep than for shallow processing during encoding. Since a recognition task succeeding the lexical decision task also indicated the level of processing effects with better recognition of deeply processed words, this enhanced positive slow wave (pSW) can be interpreted as an index of recollection (Paller et al., 1995; Paller, 2000). In the present study, ERPs of PD patients registered during a lexical decision task were analyzed as a sensitive electrophysiological measure of recollection. The “emotion-induced recognition bias” (e.g., Maratos, Allan, & Rugg, 2000; Windmann & Kutas, 2001) refers to the observation that healthy participants in a recognition test classify emotionally negative words as “old” more often than with emotionally neutral words, whether the words are in fact old or not. In other words, participants tend to adopt a more liberal response criterion to emotional than to neutral words. This decision criterion shift for emotional stimuli does not improve accuracy, but does ensure that memories for emotional stimuli, which may have a high survival value, are not easily missed or erroneously considered irrelevant. In ERPs, the emotion-induced recognition bias is reflected in reduced old/new effects at prefrontal recording sites in the time range between 300 and 500 ms (Windmann & Kutas, 2001). Recently, Windmann, Sakhavat, & Kutas (2002) researched the emotion-induced recognition bias in PD patients. During the study phase, negative and neutral words were presented to PD patients and healthy controls, and both groups were instructed to memorize the words. The recognition test revealed, for both groups, higher hit and false alarm rates for negative compared to neutral words, but no group differences or accuracy measure differences between negative and neutral words. This response pattern reflects an emotion-induced recognition bias with a response bias in favor of negative words, but no PD-specific memory bias. Simultaneously, the emotive connotation of words systematically influenced control subjects’ ERPs between 300 and 500 ms at prefrontal sites, and this effect was not observable in PD patients. While the normal behavioral results of PD patients were not expected by the authors and are at variance with previous reports of an explicit memory bias in PD for threat stimuli (see Coles & Heimberg, 2002), the ERP findings confirmed the authors’ expectations of a dysfunctional inhibitory modulation of affective information processing in PD. One reason why no memory bias for PD patients was found may be the fact that word stimuli with a general negative connotation were compared to neutral stimuli, although there are no studies showing that general negative and PD-specific threat stimuli have differential effects on memory. The present study further elucidated the nature of the memory bias in PD patients. The experimental design mirrored Paller et al. (1995) with an incidental learning task followed by lexical decision and recognition tasks which assessed implicit and explicit memory, respectively. ERPs registered during the lexical decision task were analyzed as an additional measure of recollective processes (Paller et al., 1995). Depth of processing during encoding was varied through instructions (syntactic vs. orthographic processing), and threatening body-related and neutral words were presented to PD patients and healthy controls. Since the literature is equivocal regarding an implicit memory bias in PD, we had no clear expectancy for the behavioral lexical decision measures. However, ERPs during lexical decisions should reflect recollection processes ( Paller et al., 1995). Therefore, we expected more positive ERPs triggered by threat stimuli in PD patients than in healthy participants, and more positive ERPs for syntactically compared to orthographically processed stimuli. Several previous studies reported an explicit memory bias in panic patients. For the recognition task, therefore, we expected higher discrimination indices for panic-related words in PD patients compared to healthy participants, but no intergroup differences in discrimination indices for neutral words. These effects were expected to be especially strong for words that were deeply processed during encoding.