جانبداری حافظه ضمنی و اضطراب صفت: تجزیه و تحلیل فیزیولوژیک روانی

The effect of threatening words and anxiety upon implicit memory performance was investigated. It was predicted that anxious individuals would show a bias to threat-related material. In addition, psychophysiological measures were obtained to assess the attentional and encoding processes that might underlie this cognitive bias. Forty participants were equally allocated to high and low trait anxious groups, according to pre-determined cut-offs. All participants were exposed to threat and non-threat words and following a filler task, were asked to complete primed and unprimed wordstems. Implicit memory performance was assessed in terms of accuracy and reaction time for completion. Heart rate and electrodermal responses were measured. Results demonstrated initial increased cardiac deceleration to threat stimuli, subsequent cardiac acceleration to non-threat stimuli, and an implicit memory bias to non-threat material by all participants. These findings are discussed in relation to the ‘vigilance-avoidance’ model of attention to threat stimuli.

Contemporary models of emotional disorders (e.g. Williams et al., 1988 and Williams et al., 1997) have emphasised dissociations in cognitive processes subsuming attention and memory within anxiety and depression. Essentially, attentional biases towards affective material, such as threatening information, are said to be more prevalent within anxiety compared to depression. In contrast, biases in the mood congruent recall of past events, such as autobiographical or explicit memory tasks, are said to characterise depressive disorders. Measures of implicit memory, on the other hand, are thought to reflect the pre-attentive integrative processes that underlie attentional biases demonstrated in anxiety. Consequently, Williams et al., 1988 and Williams et al., 1997 proposed that anxious individuals would also demonstrate implicit memory biases for threatening information. Unfortunately, studies that have examined emotional biases in implicit memory tasks have reported somewhat equivocal findings in both clinical and trait anxious samples (Eysenck and Byrne, 1994, Lang and Craske, 1997, Mathews et al., 1989, Richards and French, 1991, Bradley et al., 1994, Mathews et al., 1995 and Nugent and Mineka, 1994). Indeed, Russo et al. (1999) recently reviewed the literature and, together with their own studies of implicit memory, concluded that there was no substantive evidence to support an anxiety bias within implicit memory. Given the importance of implicit memory tasks in providing a theoretical link between attentional and perceptual encoding explanations of emotionally related cognitive biases, the present study set out to re-examine the processes underlying these reported biases in word completion tasks. Explicitly, the study sought to relate attentional performance during the encoding (priming) phase of a wordstem completion task to subsequent performance on the task itself. It was predicted that implicit memory would be greater for threat related stimuli, compared with non-threat related material within high trait anxious individuals and that such an effect would be related to greater attention for threat stimuli during the priming phase of the wordstem completion task. Todate such an approach has seldom been adopted which is understandable given the complexities of combining attentional bias paradigms (i.e. Stroop or Dot-probe) alongside word completion tasks. In recognition of these procedural difficulties, the current study adopted a different approach, relying upon a psychophysiological analysis of the relevant experimental paradigms. It will be argued that psychophysiological responses quantified during the encoding phase should provide measures of both perceptual and emotional processing which could be related to subsequent task performance. The rationale for such an approach is elaborated below. It is surprising that given the recent emphasis on information processing models of cognitive bias in emotional disorders, few studies have employed psychophysiological measures. At a conceptual level, the attentional shifts described within the cognitive-emotional literature bear many similarities to current theoretical positions regarding psychophysiological models of orienting and attention (Graham, 1997). Although these traditional information processing approaches within psychophysiology have tended to focus on the physical attributes of stimuli, these frameworks have now been extended to affective material. For example, Cook and Turpin (1997) describe different pre-attentive and evaluative mechanisms underlying defensive and startle responding. Similarly, Öhman (1997) has proposed a pre-attentive orienting mechanism specific to evolutionary important threat cues. Essentially, these models predict that novel and significant stimuli give rise to orienting responses, as indexed by heart rate responding and electrodermal activity. More specifically, the degree of attention directed towards these stimuli would primarily be determined by their novelty and significance and be indexed by the extent of heart rate deceleration and the amplitude of the electrodermal response. It is also claimed that threat related stimuli are initially detected at a pre-attentive stage and thereafter a variety of attentional responses ensue. Firstly, abrupt and sudden stimuli would give rise to a powerful interrupt system, as indexed by startle responding. Secondly, threatening stimuli would give rise either to further orienting responses associated with additional perceptual processing, or to defensive-type responding related to sensory-motor avoidance and escape. The above attentional responses are to be identified by early and late bi-directional components of the heart rate response (Cook and Turpin, 1997). We would argue therefore, that these psychophysiological attentional responses parallel the different types of attentional bias described in the clinical-emotional literature. Whilst orienting to threat related material is synonymous with an attentional bias towards threat, a defense response reflects an attentional shift away. Indeed, a study by Thayer et al. (2000) adopted a similar interpretation of cardiac responses in their investigation of attentional responses to threat related semantic stimuli. In addition to psychophysiological measures being employed as indices of attention, they are also thought to reflect subsequent processing and elaboration of information. Jennings (1986) argued that cardiac responding may be differentiated into an initial decelerative attentional component which is followed by a subsequent accelerative component indicative of further cognitive processing. It has also been suggested that electrodermal responding, as well as reflecting the allocation of attention to novel stimuli as part of the orienting response, may reflect subsequent encoding or rehearsal processes related to memory (cf. Craik and Blankstein, 1975, Kleinsmith and Kaplan, 1963, Maltzman et al., 1966 and Siddle and Packer, 1991). For a review of the relationship between autonomic responding and memory see Jennings (1986). The present study was designed to re-examine implicit memory biases towards threat using psychophysiological measures as a tool for studying the nature of attentional and encoding processes that might underlie such biases. A major aim of the study therefore, was to demonstrate the viability, or otherwise, of combining cognitive paradigms with autonomic indices of attention. Given that the current investigation involved the novel combining of essentially two disparate paradigms, we decided to adopt an implicit memory paradigm on account that it is a well-researched experimental design within the cognitive literature of anxiety. Specifically, we chose to replicate, as precisely as possible, a published study of wordstem completion (e.g. Richards and French, 1991) that had demonstrated trait anxiety effects on implicit memory using an undergraduate student sample. The paradigm involved an encoding phase followed by a test phase which consisted of presenting wordstems that related either to the words presented previously (primed) or to unseen words (unprimed). Participants had to complete the wordstem with the first word that came to mind. Two sets of semantic material (threatening and non-threatening) were used from previously published wordlists (Mathews et al., 1989) and presented to groups of high and low trait anxious undergraduate students. Measures of electrodermal activity and heart rate were obtained throughout the encoding phase. In order to avoid response competition between successive physiological responses, the inter-stimulus-interval was varied between 25 and 35 s. It should be noted that this led to a much slower speed of presentation compared with that usually employed in these studies and also consequently extended the duration of the study. After the encoding phase, each participant completed a filler task, followed by the wordstem completion task. Psychophysiological measures were not recorded during the test phase of the task due to the confounding effect of physical movements made whilst completing the wordstem completion task. In order to emphasise the perceptual encoding required within the implicit memory task, participants were not asked to rehearse for a later explicit recall or recognition task. Moreover, although a measure of recognition might have been useful, the duration of the study mitigated against the inclusion of a further test phase. To obtain additional support for an implicit memory bias, the reaction time taken by participants to identify words for each of the wordstems was recorded. Finally, participants were asked to rate the threat content of each of the words presented during the encoding phase. Based on previous studies, it was predicted that high trait, relative to low trait, anxious participants would complete proportionately more threat related primed wordstems than non-threat related primed wordstems. Similarly, the reaction time would be shorter for primed threat related wordstem completion. It was also hypothesised that threat words would elicit larger electrodermal and cardiovascular responses than non-threat words. In addition, heart rate responding would be related to subsequent performance on the wordstem completion task. Greater wordstem completion for threat material in the high anxious group would be associated with greater initial heart rate deceleration, followed by greater subsequent acceleration. The decelerative component would reflect greater pre-attentive processing of the threat stimulus whereas the accelerative component might reflect subsequent evaluative or elaborative processing of this material.

The present study has demonstrated the feasibility of incorporating autonomic psychophysiological measures into the study of cognitive-emotional biases within memory. For individuals who are not clinically anxious, both performance and psychophysiological measures indicate that non-threat relative to threat material is more efficiently encoded into memory, as assessed by an implicit memory paradigm. We also speculate that non-anxious individuals initially detect threat-related material but their bias for non-threat develops as the possible result of subsequent deployment of attention away from negative stimuli.